A Three‐dimensional Diagnostic Diagram for Seyfert 2 Galaxies: Probing X‐Ray Absorption and Compton Thickness

We present a study of the intrinsic absorption lines in the ultraviolet spectra of Seyfert 1 galaxies. The study is based on spectra from the Hubble Space Telescope and includes the Seyfert 1 galaxies observed with the Faint Object Spectrograph and Goddard High-Resolution Spectrograph at spectral resolutions of λ/Δλ ≈1000-20,000 with good signal-to-noise ratios. We find that the fraction of Seyfert 1 galaxies that show intrinsic absorption associated with their active nuclei is more than one-half (10 of 17), which is much higher than previous estimates (3%-10%) based on IUE data. There is a one-to-one correspondence between Seyfert galaxies that show intrinsic UV absorption and X-ray warm absorbers, indicating that these two phenomena are related. Although our sample is not complete, we conclude that intrinsic absorption represents an important component that needs to be integrated into our overall physical picture of active galaxies. The intrinsic UV absorption is generally characterized by high ionization: C IV and N V are seen in all 10 Seyfert galaxies with detected absorption (in addition to Lyα), whereas Si IV is present in only four of these Seyfert galaxies, and Mg II absorption is detected only in NGC 4151. The absorption lines are blueshifted (or in a few cases at rest) with respect to the narrow emission lines, indicating that the absorbing gas is undergoing net radial outflow. At high resolution, the absorption often splits into distinct kinematic components that show a wide range in widths (20-400 km s-1 FWHM), indicating macroscopic motions (e.g., radial velocity subcomponents or turbulence) within a component. The strong absorption components have cores that are much deeper than the continuum flux levels, indicating that the regions responsible for these components lie completely outside of the broad emission-line regions. Additional information on the covering factors and column densities can be derived from the absorption profiles in the high-resolution spectra. The covering factor of the absorbing gas in the line of sight, relative to the total underlying emission, is Clos ≥ 0.86, on average. The global covering factor, which is the fraction of emission intercepted by the absorber averaged over all lines of sight, is Cglobal ≥ 0.5. Thus, structures covering large solid angles as seen by the central continuum source (e.g., spherical shells, sheets, or cones with large opening angles) are required. The individual absorptioncomponents show a wide range in C IV column densities (0.1-14 × 1014 cm-2), and the ratio of N V to C IV column density varies significantly from one absorption component to the next, even in the same Seyfert galaxy. Thus, the intrinsic absorption in a Seyfert 1 galaxy is typically comprised of distinct kinematic components that are characterized by a range in physical conditions (e.g., ionization parameter and hydrogen column density). Finally, we show evidence for extreme variability in the intrinsic absorption lines of NGC 3783. In addition to our earlier report of the appearance of a C IV absorption doublet at -560 km s-1 (relative to the emission lines) over 11 months, we have detected the appearance of another C IV doublet at -1420 km s-1 over 15 months. On the other hand, the C IV absorption lines of NGC 3516 and NGC 4151 were very stable over periods of 6 months and 4 years, respectively. Monitoring observations of individual Seyfert galaxies at higher time resolution are needed to distinguish between different sources of variability (variable ionization, motion of gas across the line of sight) and to determine the densities and radial locations of the absorption components.

We study the X-ray absorption of a complete sample of 99 bright Swift gamma-ray bursts. Over the last few years, a strong correlation between the intrinsic X-ray absorbing column density (N_H(z)) and the redshift was found. This absorption excess in high-z GRBs is now thought to be due to the overlooked contribution of the absorption along the intergalactic medium, by means of both intervening objects and the diffuse warm-hot intergalactic medium along the line of sight. In this work we neglect the absorption along the IGM, because our purpose is to study the eventual effect of a radical change in the Galactic absorption model on the N_H(z) distribution. Therefore, we derive the intrinsic absorbing column densities using two different Galactic absorption models, the Leiden Argentine Bonn HI survey and the more recent model including molecular hydrogen. We find that, if on the one hand the new Galactic model considerably affects the single column density values, on the other hand there is no drastic change in the distribution as a whole. It becomes clear that the contribution of Galactic column densities alone, no matter how improved, is not sufficient to change the observed general trend and it has to be considered as a second-order correction. The cosmological increase of N_H(z) as a function of redshift persists and, in order to explain the observed distribution, it is necessary to include the contribution of both the diffuse intergalactic medium and the intervening systems along the line of sight of the GRBs.

Nearby active galactic nuclei were diagnosed in the X-ray and mid-to-far infrared wavelengths with Monitor of All-sky X-ray Image (MAXI) and the Japanese infrared observatory AKARI, respectively. One hundred of the X-ray sources listed in the second release of the MAXI all-sky X-ray source catalog are currently identified as non-blazar-type active galactic nuclei. These include 95 Seyfert galaxies and 5 quasars, and they are composed of 73 type-1 and 27 type-2 objects. The AKARI all-sky survey point source catalog was searched for their mid- and far-infrared counterparts at 9, 18, and 90 μm. As a result, 69 Seyfert galaxies in the MAXI catalog (48 type-1 and 21 type-2) were found to be detected with AKARI. The X-ray (3–4 keV and 4–10 keV) and infrared luminosities of these objects were investigated, together with their color information. Adopting the canonical photon index, Γ = 1.9, of the intrinsic X-ray spectrum of the Seyfert galaxies, the X-ray hardness ratio between the 3–4 and 4–10 keV ranges derived with MAXI was roughly converted into the absorption column density. After the X-ray luminosity was corrected for absorption from the estimated column density, the well-known X-ray-to-infrared luminosity correlation was confirmed, at least in the Compton-thin regime. In contrast, NGC 1365, the only Compton-thick object in the MAXI catalog, was found to deviate from the correlation toward a significantly lower X-ray luminosity by nearly an order of magnitude. It was verified that the relation between the X-ray hardness below 10 keV and X-ray-to-infrared color acts as an effective tool to pick up Compton-thick objects. The difference in the infrared colors between the type-1 and type-2 Seyfert galaxies and its physical implication on the classification and unification of active galactic nuclei are briefly discussed.

view Abstract Citations (33) References (31) Co-Reads Similar Papers Volume Content Graphics Metrics Export Citation NASA/ADS The Optical Counterpart to the Luminous X-Ray Supernova Remnant in NGC 6946 Blair, William P. ; Fesen, Robert A. Abstract We identify a high surface brightness optical supernova remnant (SNR) in NGC 6946 that is coincident with a bright soft X-ray point source detected in a ROSAT PSPC image of this galaxy (Schlegel 1994). Optical CCD/interference filter images and spectra show the object to have an observed Hα flux of 1.9 x 10^-14^ ergs cm^-2^ s^-1^, corresponding to an intrinsic Hα luminosity of 2 x 10^38^ (d/5.1 Mpc)^2^ ergs s^- 1^. A moderate-resolution optical spectrum shows normal SNR line emissions, with no high velocities (V <400 km s^-1^) or relative line strengths indicative of enriched heavy element abundances. This suggests a relatively old SNR, a conclusion supported by the presence of an emission nebula at its position on photographic plates dating back to at least 1921. With both an optical line flux and soft X-ray flux exceeding those of other luminous extragalactic SNRs, this object may have an optical luminosity near the maximum limit for SNRs. Assuming current flux levels, we estimate the SNR's age to be <= 3500 yr and suggest its unusual luminosity is due to both its expansion into dense surroundings and a relatively low column density along our line of sight. Publication: The Astrophysical Journal Pub Date: April 1994 DOI: 10.1086/187285 Bibcode: 1994ApJ...424L.103B Keywords: Balmer Series; Data Reduction; H Alpha Line; Light (Visible Radiation); Optical Emission Spectroscopy; Supernova Remnants; X Ray Sources; Charge Coupled Devices; Line Of Sight; Luminosity; Photographic Plates; Rosat Mission; Astrophysics; GALAXIES: INDIVIDUAL NGC NUMBER: NGC 6946; ISM: SUPERNOVA REMNANTS full text sources ADS | data products SIMBAD (6) NED (1) HEASARC (1)

We report here results from detailed timing and spectral studies of the high mass X-ray binary pulsar 4U 1538-52 over several binary periods using observations made with the Rossi X-ray Timing Explorer (RXTE) and BeppoSAX satellites. Pulse timing analysis with the 2003 RXTE data over two binary orbits confirms an eccentric orbit of the system. Combining the orbitial parameters determined from this observation with the earlier measurements we did not find any evidence of orbital decay in this X-ray binary. We have carried out orbital phase resolved spectroscopy to measure changes in the spectral parameters with orbital phase, particularly the absorption column density and the iron line flux. The RXTE-PCA spectra in the 3–20 keV energy range were fitted ∼6.4 keV, whereas the BeppoSAX spectra needed only a power law and Gaussian emission line at ∼6.4 keV in the restricted energy range of 0.3–10.0 keV. An absorption along the line of sight was included for both the RXTE and BeppoSAX data. The variation of the free spectral parameters over the binary orbit was investigated and we found that the variation of the column density of absorbing material in the line of sight with orbital phase is in reasonable agreement with a simple model of a spherically symmetric stellar wind from the companion star.

We report simulations of emission-line morphologies and results of a large imaging survey of a complete sample of Seyfert galaxies in early- type hosts in the emission lines of [O III] λ5007 and Hα + [N III] λλ6548, 6583 and the nearby continua. These simulations and observations are compared in order to test models in which the apparent Seyfert class depends on the orientation of the active nucleus with respect to the observer. In unified models, the active nucleus is surrounded by a thick, dusty gaseous torus. The nuclear ionizing radiation is then expected to escape in oppositely directed cones along the axis of the torus. We have, therefore, performed simulations in which ambient gas is ionized by a bi-conical radiation field. When the ambient gas is distributed in a sphere or spheroid, V- shaped projections are expected whenever our line of sight is outside the cones of ionizing radiation. For line of sights within the cones, "halo"-like morphologies are produced. For such gas distributions, the measured opening angle of the V-shaped morphology is equal to or larger than the true opening angle of the radiation field. On the other hand, when the gas is distributed in a thin disk, V-shaped projections are produced the majority of the time, even when our line of sight is within the photon cone. The measured opening angle in the case is usually smaller than the true opening angle of the photon cone. For both spherical and thin disk gas distributions, the projected linear extent of the ionized gas tends to be smaller for lines of sight within the cones. From the emission-line images, we find that extended emission-line gas is very common in Seyfert galaxies, with ~80% of the galaxies showing extension in our [O III] λ5007 images and ~100% showing extension in Hα + [N II]. The percentage of galaxies extended in these lines is comparable for Seyfert 1s and 2s. There is a strong correlation between the line fluxes of the unresolved core and the extended emission, suggesting the extended emission is ionized by the same source that ionizes the nuclear gas. There is also a correlation between emission- line extent and luminosity. The total [O III] λ5007 luminosities of the Seyfert 1s in our sample are similar to those of the Seyfert 2s, while the total Hα + [N II] emission is larger in Seyfert 1s than Seyfert 2s. This result is not surprising since the broad component to the Hα line contributes a significant amount of flux in the Seyfert 1s. In order to compare the extended emission-line luminosities of the two Seyfert types, we have subtracted an unresolved nuclear component from the images. We find a trend for the extended emission in both [O III] λ5007 and Hα + [N II] to be more luminous in the Seyfert 1s than the Seyfert 2s. However, this trend is of only marginal significance in the complete sample. In general, it is difficult, if not impossible, to distinguish a Seyfert 1 galaxy from a Seyfert 2 galaxy based only on its emission-line characteristics. While linear, bipolar, or V-shaped morphologies are seen in some objects, for one-third of the sample the emission-line structures are "halo "-like. These "halo" morphologies are not expected for Seyfert 2s in the unified model. Many of the "halo" Seyfert 2s do display linear or V-shaped structures in their excitation maps (i.e., distributions of [O III]/Hα + [N II]), so they may still be consistent with orientation - dependent models. However, the emission-line morphologies of at least a few Seyfert 1s appear inconsistent with those expected in the simplest unified schemes. In our complete sample, the extent of the [O III] λ5007 emission at a given surface brightness is larger in the Seyfert 2s than the Seyfert 1s. However, this difference disappears if the two most extended Seyfert 2s are omitted. We also find evidence for structure in many of the continuum color maps. In ~40% of the galaxies with color maps, red (V - R &gt; 1.0) features are found. The majority of these red regions are unresolved structures located at the nucleus. These features are found in galaxies of all inclinations, but occur almost exclusively in type 2 Seyfert galaxies. These characteristics are consistent with reddening by ~100 pc scale, dusty tori. A less likely alternative is that these structures represent a very red stellar population (later than K5). Diffuse blue (V - R &lt; 0.5) features are seen in some Seyfert 2 galaxies. These blue excesses tend to be spatially coincident with the high-excitation gas, suggesting an origin related to the nuclear activity. The blue excesses could originate from a number of different processes, including scattered nuclear light or an extended nonstellar continuum associated with the ionized gas (e.g., shock waves).

Context. Vela X−1 is one of the first few high-mass X-ray binary (HMXB) pulsars to be discovered. In HMXBs with pulsars such as Vela X−1, the companion’s stellar wind is significantly affected by ionisation due to X-rays from the compact object. An isotopic stellar wind model alone cannot explain the orbital variation in the absorption column density in Vela X−1. A model describing a stream-like photoionisation wake trailing the neutron star has been previously implemented to explain the observed orbital column density variation. Aims. We investigated the variability of the circumbinary environment at different intensity levels of the Vela X−1 and used a model similar to the above-mentioned stream-like photoionisation wake to explain the asymmetric absorption column density present in the source. Methods. The 2.0−20.0 keV MAXI/GSC spectrum was well modelled with a comptonised continuum emission absorbed by local and interstellar material. We used ∼13 years of MAXI/GSC data to constrain the variations in the absorption column density in Vela X−1 obtained from orbital-phase resolved and intensity-and-orbital-phase resolved spectral analysis. Results. The long-term light curve of Vela X−1 shows orbit-to-orbit intensity level variations without any apparent super-orbital periodicity. The orbital-phase resolved spectroscopy in multiple intensity levels reveals asymmetric variation in absorption column density changes across the intensity levels. Conclusions. We confirm that the orbital variation in the absorption column density in Vela X−1 cannot be modelled with a smooth stellar wind alone using ∼13 years of MAXI/GSC data. It requires an additional component, such as a photoionisation wake or an accretion wake. The wake structure is found to be present across different intensity levels of the source, and the geometry of the wake depends on the intensity level. The long-duration MAXI/GSC data allowed us to vary different wake parameters to obtain the best-fit stellar wind parameters for the time-averaged intensity. These best-fit parameters closely reproduce the observed orbital variations in the absorption column density for different intensity levels of the source.

We remark on the utility of an observational relation between the absorption column density in excess of the Galactic absorption column density, ΔNH = NH,fit - NH,gal, and redshift, z, determined from all 55 Swift-observed long bursts with spectroscopic redshifts as of 2006 December. The absorption column densities, NH,fit, are determined from power-law fits to the X-ray spectra with the absorption column density left as a free parameter. We find that higher excess absorption column densities with ΔNH > 2 × 1021 cm-2 are only present in bursts with redshifts z < 2. Low absorption column densities with ΔNH < 1 × 1021 cm-2 appear preferentially in high-redshift bursts. Our interpretation is that this relation between redshift and excess column density is an observational effect resulting from the shift of the source rest-frame energy range below 1 keV out of the X-Ray Telescope observable energy range for high-redshift bursts. We find a clear anticorrelation between ΔNH and z that can be used to estimate the range of the maximum redshift of an afterglow. A critical application of our finding is that rapid X-ray observations can be used to optimize the instrumentation used for ground-based optical/near-IR follow-up observations. Ground-based spectroscopic redshift measurements of as many bursts as possible are crucial for gamma-ray burst science.

We present a broad-band (~0.3-70 keV) spectral and temporal analysis of NuSTAR observations of the luminous infrared galaxy NGC 6240, combined with archival Chandra, XMM-Newton and BeppoSAX data. NGC 6240 is a galaxy in a relatively early merger state with two distinct nuclei separated by ~1."5. Previous Chandra observations have resolved the two nuclei, showing that they are both active and obscured by Compton-thick material. Although they cannot be resolved by NuSTAR, thanks to the unprecedented quality of the NuSTAR data at energies >10 keV, we clearly detect, for the first time, both the primary and the reflection continuum components. The NuSTAR hard X-ray spectrum is dominated by the primary continuum piercing through an absorbing column density which is mildly optically thick to Compton scattering (tau ~ 1.2, N_H ~ 1.5 x 10^(24) cm^-2). We detect moderate hard X-ray (> 10 keV) flux variability up to 20% on short (15-20 ksec) timescales. The amplitude of the variability is maximum at ~30 keV and is likely to originate from the primary continuum of the southern nucleus. Nevertheless, the mean hard X-ray flux on longer timescales (years) is relatively constant. Moreover, the two nuclei remain Compton-thick, although we find evidence of variability of the material along the line of sight with column densities N_H <~ 2 x 10^(23) cm-2 over long (~3-15 years) timescales. The observed X-ray emission in the NuSTAR energy range is fully consistent with the sum of the best-fit models of the spatially resolved Chandra spectra of the two nuclei.

RW Aur is a binary system composed of two young, low-mass stars. The primary, RW Aur A, has undergone visual dimming events (ΔV = 2–3 mag) in 2011, 2014–16, and 2017–2018. Visual and IR observations indicate a gray absorber that moved into the line of sight. This dimming is also associated with changes in the outflow. In 2017, when the optical brightness was almost 2 mag below the long-term average, we triggered a Chandra observation to measure the absorbing column density N H and to constrain dust properties and the gas-to-dust ratio of the absorber. In 2017, the X-ray spectrum is more absorbed than it was in the optically bright state ( ) and shows significantly more hot plasma than in X-ray observations taken before. Furthermore, a new emission feature at 6.63 ± 0.02 keV (statistic) ±0.02 keV (systematic) appeared, indicating an Fe abundance an order of magnitude above solar, in contrast with previous sub-solar Fe abundance measurements. Comparing X-ray absorbing column density N H and optical extinction A V , we find that either the gas-to-dust ratio in the absorber is orders of magnitude higher than in the ISM, or the absorber has undergone significant dust evolution. Given the high column density coupled with changes in the X-ray spectral shape, this absorber is probably located in the inner disk. We speculate that a breakup of planetesimals or a terrestrial planet could supply large grains, causing gray absorption; some of these grains would be accreted and enrich the stellar corona with iron, which could explain the inferred high abundance.

The X-ray transient source EP240309a/EP,J115415.8-501810 was first detected by the Wide-field X-ray Telescope on board the Einstein Probe (EP) instrument during the commissioning phase. Subsequent optical observations confirmed it as a cataclysmic variable of the intermediate polar type with a 238.2,s spinning white dwarf on a ∼3.76,hr orbit. We report the source discovery and follow-up studies carried out with the EP's Follow-up X-ray Telescope. A periodic variation of 231,s was detected in the 0.3-2,keV band, while no obvious pulsation appeared in the 2-10,keV band. The spectral analysis showed that the X-ray emission could be described by an absorbed bremsstrahlung model with kT The partial covering absorption, with a hydrogen column density N_H = 2.0times 10^ ̊m cm^ and covering fraction of around 0.9, is much higher than the interstellar absorption along the line of sight. According to the distance of d = 309.5,pc obtained from the Gaia parallax, we estimated that the luminosity of this source in the 0.3-10,keV range is ∼ 2 erg,s^-1. In addition, a phase-resolved spectral analysis revealed that the detected periodic variation is mainly caused by the change in the absorption column density. In this scenario, the spin modulation arises due to absorption from the pre-shock accretion flow of the X-ray emitting pole, while the optical radiation is modulated at the orbital side band (ω_ ̊m spin - Ω_ ̊m orbit ) due to the reprocessing in regions within the binary system. Due to this unusual transient behaviour for an intermediate polar, we also searched for radio signals similar to those observed in the new class of long period transients. We derived upper limits with ASKAP of 200--300,μJy,beam^-1 in the range 800--1500 MHz and with MWA of 40--90,mJy,beam^-1 in the range 80--300 MHz.

We systematically analyzed the high-quality Suzaku data of 88 Seyfert galaxies. We obtained a clear relation between the absorption column density and the equivalent width of the 6.4 keV line above 10$^{23}$ cm$^{-2}$, suggesting a wide-ranging column density of $10^{23-24.5}$ cm$^{-2}$ with a similar solid and a Fe abundance of 0.7--1.3 solar for Seyfert 2 galaxies. The EW of the 6.4 keV line for Seyfert 1 galaxies are typically 40--120 eV, suggesting the existence of Compton-thick matter like the torus with a column density of $>10^{23}$ cm$^{-2}$ and a solid angle of $(0.15-0.4)*4pi$, and no difference of neutral matter is visible between Seyfert 1 and 2 galaxies. An absorber with a lower column density of $10^{21-23}$ cm$^{-2}$ for Compton-thin Seyfert 2 galaxies is suggested to be not a torus but an interstellar medium. These constraints can be understood by the fact that the 6.4 keV line intensity ratio against the 10--50 keV flux is almost identical within a range of 2--3 in many Seyfert galaxies. Interestingly, objects exist with a low EW, 10--30 eV, of the 6.4 keV line, suggesting that those torus subtends only a small solid angle of $<0.2*4pi$. Ionized Fe-K$\alpha$ emission or absorption lines are detected from several percents of AGNs. Considering the ionization state and equivalent width, emitters and absorbers of ionized Fe-K lines can be explained by the same origin, and highly ionized matter is located at the broad line region. The rapid increase in EW of the ionized Fe-K emission lines at $N_{H}>10^{23}$ cm$^{-2}$ is found, like that of the cold material. It is found that these features seem to change for brighter objects with more than several $10^{44}$ erg/s such that the Fe-K line features become weak. We discuss this feature, together with the torus structure.

Observations obtained with the Far Ultraviolet Spectroscopic Explorer (FUSE) have been used to determine the column densities of D I, O I, and N I along seven sight lines that probe the local interstellar medium (LISM) at distances from 37 pc to 179 pc. Five of the sight lines are within the Local Bubble and two penetrate the surrounding H I wall. Reliable values of N(H I) were determined for five of the sight lines from HST data, IUE data, and published EUVE measurements. The weighted mean of DI/H I for these five sight lines is (1.52 +/- 0.08) x l0(exp -5)(1 sigma uncertainty in the mean). It is likely that the D I/H I ratio in the Local Bubble has a single value. The D I/O I ratio for the five sight lines within the Local Bubble is (3.76 +/- 0.20) x 10(esp -2). It is likely that O I column densities can serve as a proxy for H I in the Local Bubble. The weighted mean for O I/ H I for the seven FUSE sight lines is (3.03 +/- 0.21) x 10(esp -4), comparable to the weighted mean (3.43 +/- 0.15) x 10(exp -4) reported for 13 sight lines probing larger distances and higher column densities. The FUSE weighted mean of N I/ H I for five sight lines is half that reported by Meyer et al. for seven sight lines with larger distances and higher column densities. This result combined with the variability of O I/ N I (six sight lines) indicates that at the low column densities found in the LISM, nitrogen ionization balance is important. Thus, unlike O I, N I cannot be used as a proxy for H I or as a metallicity indicator in the LISM.

The {\it Far Ultraviolet Spectroscopic Explorer} ({\it FUSE}) has allowed precise determinations of the column densities of molecular hydrogen ($\Hmol$) in Galactic lines of sight with a wide range of pathlengths and extinction properties. However, survey studies of lines of sight with greater extinction have been mostly restricted to the low-$J$ states (lower total angular momentum) in which most molecular hydrogen is observed. This paper presents a survey of column densities for the molecular hydrogen in states of greater rotational excitation ($J \geq 2$) in Galactic lines of sight with $\log{\NHmol} \gtrsim 20$. This study is comprehensive through the highest excited state detectable in each line of sight. J=5 is observed in every line of sight, and we detect J=7 in four lines of sight, J=8 in one line of sight, and vibrationally excited $\Hmol$ in two lines of sight. We compared the apparent $b$-values and velocity offsets of the higher-$J$ states relative to the dominant low-$J$ states and we found no evidence of any trends that might provide insight into the formation of higher-$J$ $\Hmol$, although these results are the most affected by the limits of the {\it FUSE} resolution. We also derive excitation temperatures based on the column densities of the different states. We confirm that at least two distinct temperatures are necessary to adequately describe these lines of sight, and that more temperatures are probably necessary. Total $\Hmol$ column density is known to be correlated with other molecules; we explore if correlations vary as a function of $J$ for several molecules, most importantly CH and CH$^+$. Finally, we briefly discuss interpretations of selected lines of sight by comparing them to models computed using the Meudon PDR code.

Active galaxies are thought to be powered by the accretion of gas onto a central massive black hole. Seyfert galaxies—the most common examples of nearby active galaxies—are separated into two classes based on their emission line widths1. Seyfert 1 galaxies exhibit broad emission lines that are attributed to ionized gas within 1 pc of the black hole, whereas the spectra of Seyfert 2 galaxies show only narrower emission lines, believed to originate from a much larger region around the core. The 'unified model' for Seyfert galaxies attributes these differences to the presence of a dusty torus of dense molecular gas surrounding the black hole2: the orientation of Seyfert 2 galaxies is such that the broad-line region is obscured. The detection3 in the polarization spectrum of broad emission lines scattered into our line of sight by free electrons in NGC1068 (the prototypical Seyfert 2 galaxy) and other Seyfert 2 galaxies4–8 has strengthened this view, but all of these galaxies were subject to selection biases. Here we report the results of a systematic search for polarized broad emission lines in a well defined sample of Seyfert 2 galaxies. We show that the ability to detect scattered broad emission lines is related to the far-infrared colours, in the manner predicted by the unified model.