A method to measure the mass of damped Lyα absorber host galaxies using fluctuations in 21-cm emission

Observations of damped Lyα absorbers (DLAs) indicate that the fraction of hydrogen in its neutral form (H  i ) is significant by mass at all redshifts. This gas represents the reservoir of material that is available for star formation at late times. As a result, observational identification of the systems in which this neutral hydrogen resides is an important missing ingredient in models of galaxy formation. Precise identification of DLA host mass via traditional clustering studies is not practical owing to the small numbers of known systems being spread across sparsely distributed sightlines. However, following the completion of re-ionization, 21-cm surface brightness fluctuations will be dominated by neutral hydrogen in DLAs. No individual DLAs could be detected in 21-cm emission. Rather, observations of these fluctuations will measure the combined clustering signal from all DLAs within a large volume. We show that measurement of the spherically averaged power spectrum of 21-cm intensity fluctuations due to DLAs could be used to measure the galaxy bias for DLA host galaxies when combined with an independent measurement of the cosmological H  i mass density from quasar absorption studies. Utilizing this technique, the low-frequency arrays now under construction could measure the characteristic DLA host mass with a statistical precision as low as 0.3 dex at z ≳ 4. In addition, high signal-to-noise ratio observations of the peculiar-motion-induced anisotropy of the power spectrum would facilitate measurement of both the DLA host mass and the cosmic H  i density directly from 21-cm fluctuations. By exploiting this anisotropy, a second generation of low-frequency arrays with an order of magnitude increase in collecting area could measure the values of cosmic H  i density and DLA host mass, with uncertainties of a few per cent and a few tens of per cent, respectively.     

The impact of H i in galaxies on 21-cm intensity fluctuations during the reionization epoch

We investigate the impact of neutral hydrogen (H  i ) in galaxies on the statistics of 21-cm fluctuations using seminumerical modelling. Following the reionization of hydrogen, the H  i content of the Universe is dominated by damped absorption systems (DLAs), with a cosmic density in H  i that is observed to be constant at a level equal to ∼2 per cent of the cosmic baryon density from   z ∼ 1  to   z ∼ 5  . We show that extrapolation of this constant fraction into the reionization epoch results in a reduction in the amplitude of 21-cm fluctuations over a range of spatial scales. We further find that consideration of H  i in galaxies/DLAs reduces the prominence of the H  ii region induced shoulder in the 21-cm power spectrum (PS), and hence modifies the scale dependence of 21-cm fluctuations. We also estimate the 21-cm–galaxy cross PS and show that the cross PS changes sign on scales corresponding to the H  ii regions. From consideration of the sensitivity for forthcoming low-frequency arrays, we find that the effects of H  i in galaxies/DLAs on the statistics of 21-cm fluctuations will be significant with respect to the precision of a PS or cross PS measurement. In addition, since overdense regions are reionized first we demonstrate that the cross-correlation between galaxies and 21-cm emission changes sign at the end of the reionization era, providing an alternative avenue to pinpoint the end of reionization. The sum of our analysis indicates that the H  i content of the galaxies that reionize the universe will need to be considered in detailed modelling of the 21-cm intensity PS in order to correctly interpret measurements from forthcoming low-frequency arrays.     

The expected detection of dust emission from high-redshift Lyman α galaxies

Recent results have shown that a substantial fraction of high-redshift Lyman α (Lyα) galaxies contain considerable amounts of dust. This implies that Lyα galaxies are not primordial, as has been thought in the past. However, this dust has not been directly detected in emission; rather it has been inferred based on extinction estimates from rest-frame ultraviolet (UV) and optical observations. This can be tricky, as both dust and old stars redden galactic spectra at the wavelengths used to infer dust. Measuring dust emission directly from these galaxies is thus a more accurate way to estimate the total dust mass, giving us real physical information on the stellar populations and interstellar medium enrichment. New generation instruments, such as the Atacama Large Millimeter Array and Sub-Millimeter Array, should be able to detect dust emission from some of these galaxies in the submillimeter. Using measurements of the UV spectral slopes, we derive far-infrared flux predictions for of a sample of  23 z ≥ 4  Lyα galaxies. We find that in only a few hours, we can detect dust emission from 39 ± 22 per cent of our Lyα galaxies. Comparing these results to those found from a sample of 21 Lyman break galaxies (LBGs), we find that LBGs are on average 60 per cent more likely to be detected than Lyα galaxies, implying that they are more dusty, and thus indicating an evolutionary difference between these objects. These observations will provide better constraints on dust in these galaxies than those derived from their UV and optical fluxes alone. Undeniable proof of dust in these galaxies could explain the larger than expected Lyα equivalent widths seen in many Lyα galaxies today.     

A spectroscopic study of IRAS F10214 + 4724

The z  = 2.286  IRAS galaxy F10214 + 4724 remains one of the most luminous galaxies in the Universe, despite its gravitational lens magnification. We present optical and near-infrared spectra of F10214 + 4724, with clear evidence for three distinct components: lines of width ∼ 1000 km s⁻¹ from a Seyfert 2 nucleus; ≲ 200 km s⁻¹ lines which are likely to be associated with star formation; and a broad (∼ 4000 km s⁻¹) C  III ] 1909-Å emission line which is blueshifted by ∼ 1000 km s⁻¹ with respect to the Seyfert 2 lines. Our study of the Seyfert 2 component leads to several new results. (i) From the double-peaked structure in the Lyα line, and the lack of Lyβ, we argue that the Lyα photons have emerged through a neutral column of N _H ∼ 2.5 × 10²⁵ m⁻², possibly located within the AGN narrow-line region, as proposed for several high-redshift radio galaxies. (ii) The resonant O  VI 1032, 1036-Å doublet (previously identified as Lyβ) is in an optically thick (1:1) ratio. At face value this implies an extreme density ( n _e ∼ 10¹⁷ m⁻³) more typical of broad-line region clouds. However, we attribute this instead to the damping wings of Lyβ from the resonant absorption. (iii) A tentative detection of He  II 1086 suggests little extinction in the rest frame ultraviolet.     

Gas properties of H ii and starburst galaxies: relation with the stellar population

We study the gas emission of galaxies with active star formation, consisting mostly of H  ii and starburst galaxies, as well as some Seyfert 2 galaxies, and determine chemical and physical parameters. The data consist of 19 high signal-to-noise ratio optical templates, a result of grouping 185 emission-line galaxy spectra. Underlying stellar population models (from Raimann et al.) were subtracted from the templates in order to isolate the pure emission component.
We analyse the distribution of these improved signal-to-noise ratio emission spectra in diagnostic diagrams and find that the H  ii templates show a smaller spread in log([O  iii ]/H β ) values than the individual galaxies, apparently as a result of the population subtraction and a better signal-to-noise ratio. We thus suggest the template sequence as a fiducial observational locus for H  ii galaxies which can be used as reference for models. The sequence of line ratios presented by the H  ii galaxies in the diagram log([O  iii ] λ 5007/H β ) versus log([N  ii ] λ 6584/H α ) is primarily owing to the gas metallicity, of which the log([N  ii ]/H α ) ratio is a direct estimator. We also study the properties of the starburst galaxies and those intermediate between H  ii and starburst galaxies, which are more metal rich and sit on more massive galaxies.
We discuss the present results in the frame of a recently proposed equivalent-width diagnostic diagram for emission-line galaxies (by Rola et al.) and conclude that the observed ranges in W ([O  ii ])/ W (H β ) and W (H β ) are mostly owing to the non-ionizing stellar population contribution. We propose that W (H β ) be used as an estimator of this contribution to the continuum, and briefly discuss implications to the cosmological use of H  ii galaxies.     

The evolution of superbubbles and the detection of Lyα in star-forming galaxies

The detection of Ly α emission in star-forming galaxies in different shapes and intensities (always smaller than predicted for case B recombination) has puzzled the astronomical community for more than a decade. Here we use two-dimensional calculations to follow the evolution of superbubbles and of the H  ii regions generated by the output of UV photons from massive stars. We show the impact caused by massive star formation in the ISM of different galaxies and we look at the conditions required to detect Ly α emission from a nuclear H  ii region, and the variety of profiles that may be expected as a function of time.     

Glimpsing through the high-redshift neutral hydrogen fog

We analyse the transmitted flux in a sample of 17 QSOs spectra at 5.74 ≤ z _em≤ 6.42 to obtain tighter constraints on the volume-averaged neutral hydrogen fraction, x _{H  i} , at z ≈ 6. We study separately the narrow transmission windows (peaks) and the wide dark portions (gaps) in the observed absorption spectra. By comparing the statistics of these spectral features with a semi-analytical model of the Lyα forest, we conclude that x _{H  i} evolves smoothly from 10^−4.4 at   z = 5.3  to 10^−4.2 at   z = 5.6  , with a robust upper limit x _{H  i} < 0.36 at   z = 6.3  . The frequency and physical sizes of the peaks imply an origin in cosmic underdense regions and/or in H  ii regions around faint quasars or galaxies. In one case (the intervening H  ii region of the faint quasar RD J1148+5253 at   z = 5.70  along the line of sight of SDSS J1148+5251 at   z = 6.42  ) the increase of the peak spectral density is explained by the first-ever detected transverse proximity effect in the H  i Lyα forest; this indicates that at least some peaks result from a locally enhanced radiation field. We then obtain a strong lower limit on the foreground QSO lifetime of t _Q > 11 Myr. The observed widths of the peaks are found to be systematically larger than the simulated ones. Reasons for such discrepancy might reside either in the photoionization equilibrium assumption or in radiative transfer effects.     

Discovery of a single faint AGN in a large sample of z > 5 Lyman break galaxies

As part of a large spectroscopic survey of   z > 5  Lyman break galaxies (LBGs), we have identified a single source which is clearly hosting an active galactic nucleus (AGN). Out of a sample of more than 50 spectroscopically confirmed R -band dropout galaxies at   z ∼ 5  and above, only J104048.6−115550.2 at   z = 5.44  shows evidence for a high ionization potential emission line indicating the presence of a hard ionizing continuum from an AGN. Like most objects in our sample the rest-frame-UV spectrum shows the UV continuum breaking across a Lyα line. Uniquely within this sample of LBGs, emission from N  v is also detected, a clear signature of AGN photoionization. The object is spatially resolved in Hubble Space Telescope ( HST ) imaging. This, and the comparatively high Lyα/N  v flux ratio indicates that the majority of the Lyα (and the UV continuum longward of it) originates from stellar photoionization, a product of the ongoing starburst in the LBG. Even without the AGN emission, this object would have been photometrically selected and spectroscopically confirmed as a Lyman break in our survey. The measured optical flux  ( I _AB= 26.1)  is therefore an upper limit to that from the AGN and is of order 100 times fainter than the majority of known quasars at these redshifts. The detection of a single object in our survey volume is consistent with the best current models of high redshift AGN luminosity function, providing a substantial fraction of such AGN is found within luminous starbursting galaxies. We discuss the cosmological implications of this discovery.     

New insights into the stellar content and physical conditions of star-forming galaxies at z= 2–3 from spectral modelling

We have used extensive libraries of model and empirical galaxy spectra [assembled, respectively, from the population synthesis code of Bruzual and Charlot and the fourth data release of the Sloan Digital Sky Survey (SDSS)] to interpret some puzzling features seen in the spectra of high-redshift star-forming galaxies. We show that a stellar He  ii  λ1640 emission line, produced in the expanding atmospheres of Of and Wolf–Rayet stars, should be detectable with an equivalent width of 0.5–1.5 Å in the integrated spectra of star-forming galaxies, provided the metallicity is greater than about half solar. Our models reproduce the strength of the He  ii  λ1640 line measured in the spectra of Lyman-break galaxies for established values of their metallicities. With better empirical calibrations in local galaxies, this spectral feature has the potential of becoming a useful diagnostic of massive star winds at high, as well as low redshifts.
We also uncover a relationship in SDSS galaxies between their location in the [O  iii ]/Hβ versus [N  ii ]/Hα diagnostic diagram (the BPT diagram) and their excess specific star formation rate relative to galaxies of similar mass. We infer that an elevated ionization parameter U is at the root of this effect, and propose that this is also the cause of the offset of high-redshift star-forming galaxies in the BPT diagram compared to local ones. We further speculate that higher electron densities and escape fractions of hydrogen ionizing photons may be the factors responsible for the systematically higher values of U in the H  ii regions of high-redshift galaxies. The impact of such differences on abundance determinations from strong nebular lines are considered and found to be relatively minor.     

The evolution of the Lyα forest effective optical depth following He ii reionization

Three independent observational studies have now detected a narrow  (Δ z ≃ 0.5)  dip centred at   z = 3.2  in the otherwise smooth redshift evolution of the Lyα forest effective optical depth. This feature has previously been interpreted as an indirect signature of rapid photoheating in the intergalactic medium (IGM) during the epoch of He  ii reionization. We examine this interpretation using a semi-analytic model of inhomogeneous He  ii reionization and high-resolution hydrodynamical simulations of the Lyα forest. We instead find that a rapid  (Δ z ≃ 0.2)  boost to the IGM temperature  (Δ T ≃ 10⁴ K)  beginning at   z = 3.4  produces a well understood and generic evolution in the Lyα effective optical depth, where a sudden reduction in the opacity is followed by a gradual, monotonic recovery driven largely by adiabatic cooling in the low-density IGM. This behaviour is inconsistent with the narrow feature in the observational data. If photoheating during He  ii reionization is instead extended over several redshift units, as recent theoretical studies suggest, then the Lyα opacity will evolve smoothly with redshift. We conclude that the sharp dip observed in the Lyα forest effective optical depth is instead most likely due to a narrow peak in the hydrogen photoionization rate around   z = 3.2  , and suggest that it may arise from the modulation of either reprocessed radiation during He  ii reionization, or the opacity of Lyman limit systems.     

The H i opacity of the intergalactic medium at redshifts 1.6 < z < 3.2

We use high-quality echelle spectra of 24 quasi-stellar objects to provide a calibrated measurement of the total amount of Lyα forest absorption (DA) over the redshift range  2.2 < z < 3.2  . Our measurement of DA excludes absorption from metal lines or the Lyα lines of Lyman-limit systems and damped Lyα systems. We use artificial spectra with realistic flux calibration errors to show that we are able to place continuum levels that are accurate to better than 1 per cent. When we combine our results with our previous results between  1.6 < z < 2.2  , we find that the redshift evolution of DA is well described over  1.6 < z < 3.2  as   A (1 + z )^γ  , where   A = 0.0062  and  γ= 2.75  . We detect no significant deviations from a smooth power-law evolution over the redshift range studied. We find less H  i absorption than expected at   z = 3  , implying that the ultraviolet background is  ∼40  per cent higher than expected. Our data appears to be consistent with an H  i ionization rate of  Γ∼ 1.4 × 10⁻¹² s⁻¹  .     

Oxygen and helium abundances in Galactic H ii regions – I. Observations

Absolute integrated line fluxes of H  ii regions have been measured using a Fabry–Perot spectrophotometer. We describe the observations and calibration procedures. Fluxes are given for 36 H  ii regions with Galactocentric distances ranging from 6.6 to 17.7 kpc. Several emission lines have been measured, mainly [O  ii ] λλ 3726 and 3629, H β , [O  iii ] λ 5007, He  i λ 5876 and H α . The very faint [O  iii ] λ 4363 line has been measured in six regions, allowing a direct determination of the electron temperature. New photometric distances have been derived based on data from the literature. A discussion of these results in terms of extinction, electron density and temperature, and oxygen and helium abundances is given in Paper II.     

The S2N2 metallicity calibrator and the abundance gradient of M 33

We introduce the log(Hα/[S  ii ]λλ6717+6731) versus log(Hα/[N  ii ]λ6583) (S2N2) diagnostic diagram as a metallicity and ionization parameter indicator for H  ii regions in external galaxies. The location of H  ii regions in the S2N2 diagram was studied both empirically and theoretically. We found that, for a wide range of metallicities, the S2N2 diagram gives single-valued results in the metallicity–ionization parameter plane. We demonstrate that the S2N2 diagram is a powerful tool to estimate metallicities of high-redshift  ( z ∼ 2)  H  ii galaxies. Finally, we derive the metallicity for 76 H  ii regions in M 33 from the S2N2 diagram and calculate an O/H abundance gradient for this galaxy of −0.05 (±0.01) dex kpc⁻¹.     

On the polarization of Hα scattered by neutral hydrogen in active galactic nuclei

Raman scattering by atomic hydrogen converts the UV continuum around Lyβ into optical continuum around Hα, and the basic atomic physics has been discussed in several works on symbiotic stars. We propose that the same process may operate in active galactic nuclei (AGN) and calculate the linear polarization of the broad emission lines Raman-scattered by high-column neutral hydrogen component. The conversion efficiency of the Raman scattering process is discussed and the expected scattered flux is computed using the spectral energy distribution of an AGN given by a typical power law. The high-column H  i component in AGN is suggested by many observations, encompassing the radio through UV and X-ray ranges.   When neutral hydrogen component with a column density ∼10²² cm⁻² is present around the active nucleus, it is found that the scattered Hα is characterized by a very broad width ∼20 000 km s⁻¹ and that the strength of the polarized flux is comparable to that of the electron-scattered flux expected from a conventional unified model of narrow-line AGN. The width of the scattered flux is mainly determined by the column density of the neutral scatterers where the total scattering optical depth becomes of order unity. The asymmetry in the Raman scattering cross-section around Lyβ introduces red asymmetric polarized profiles around Hα. The effects of the blended Lyβ and O  vi 1034 doublet are also investigated.   We briefly discuss the spectropolarimetric observations performed on the Seyfert galaxy IRAS 110548-1131 and the narrow line radio galaxy Cyg A. Several predictions regarding the scattering by the high-column neutral hydrogen component in AGN are discussed.     

The local environment of H ii galaxies

We have carried out an investigation of the environments of low redshift H  ii galaxies by cross-correlating their positions on the sky with those of faint field galaxies in the Automatic Plate Measuring Machine (APM) catalogues. We address the question of whether violent star formation in H  ii galaxies is induced by low-mass companions by statistically estimating the mean space density of galaxies around them. We argue that even if low-mass companions were mainly intergalactic H  i clouds, their optical counterparts should be detectable at faint limits of the APM scans.
A significantly positive signal is detected for the H  ii galaxy–APM galaxy angular cross-correlation function, but the amplitude is poorly determined. The projected cross-correlation function has a higher signal-to-noise ratio, and suggests that the amplitude is slightly lower than for normal field galaxies. This implies that these bursting dwarf galaxies inhabit slightly lower density environments than those of normal field galaxies, consistent with other studies of emission-line galaxies. This suggests that in these dwarf starburst galaxies, star formation is not always triggered by tidal interactions, and a significant fraction must have a different origin.     

Physical effects on the Lyα forest flux power spectrum: damping wings, ionizing radiation fluctuations and galactic winds

We explore several physical effects on the power spectrum of the Lyα forest transmitted flux. The effects we investigate here are not usually part of hydrodynamic simulations and so need to be estimated separately. The most important effect is that of high column density absorbers with damping wings, which add power on large scales. We compute their effect using the observational constraints on their abundance as a function of column density. Ignoring their effect leads to an underestimation of the slope of the linear theory power spectrum. The second effect we investigate is that of fluctuations in the ionizing radiation field. For this purpose we use a very large high-resolution N -body simulation, which allows us to simulate both the fluctuations in the ionizing radiation and the small-scale Lyα forest within the same simulation. We find an enhancement of power on large scales for quasars and a suppression for galaxies. The strength of the effect rapidly increases with increasing redshift, allowing it to be uniquely identified in cases where it is significant. We develop templates that can be used to search for this effect as a function of quasar lifetime, quasar luminosity function and attenuation length. Finally, we explore the effects of galactic winds using hydrodynamic simulations. We find the wind effects on the Lyα forest power spectrum to be degenerate with parameters related to the temperature of the gas that are already marginalized over in cosmological fits. While more work is needed to conclusively exclude all possible systematic errors, our results suggest that, in the context of data analysis procedures, where parameters of the Lyα forest model are properly marginalized over, the flux power spectrum is a reliable tracer of cosmological information.     

ΛCDM predictions for galaxy protoclusters – I. The relation between galaxies, protoclusters and quasars at z∼ 6

Motivated by recent observational studies of the environment of   z ∼ 6  QSOs, we have used the Millennium Run (MR) simulations to construct a very large  (∼4°× 4°)  mock redshift survey of star-forming galaxies at   z ∼ 6  . We use this simulated survey to study the relation between density enhancements in the distribution of i ₇₇₅-dropouts and Lyα emitters, and their relation to the most massive haloes and protocluster regions at   z ∼ 6  . Our simulation predicts significant variations in surface density across the sky with some voids and filaments extending over scales of 1°, much larger than probed by current surveys. Approximately one-third of all   z ∼ 6  haloes hosting i -dropouts brighter than   z = 26.5  mag  (≈ M *_{UV, z =6})  become part of   z = 0  galaxy clusters. i -dropouts associated with protocluster regions are found in regions where the surface density is enhanced on scales ranging from a few to several tens of arcminutes on the sky. We analyse two structures of i -dropouts and Lyα emitters observed with the Subaru Telescope and show that these structures must be the seeds of massive clusters in formation. In striking contrast, six   z ∼ 6  QSO fields observed with Hubble Space Telescope show no significant enhancements in their i ₇₇₅-dropout number counts. With the present data, we cannot rule out the QSOs being hosted by the most massive haloes. However, neither can we confirm this widely used assumption. We conclude by giving detailed recommendations for the interpretation and planning of observations by current and future ground- and space-based instruments that will shed new light on questions related to the large-scale structure at   z ∼ 6  .     

New abundance determinations in z < 1.5 QSO absorbers: seven sub-DLAs and one DLA

We present chemical abundance measurements from high-resolution observations of seven subdamped Lyα (sub-DLA) absorbers and one DLA system at   z < 1.5  . Three of these objects have high metallicity, with near or supersolar Zn abundance. Grids of cloudy models for each system were constructed to look for possible ionization effects in these systems. For the systems in which we could constrain the ionization parameter, we find that the ionization corrections as predicted by the cloudy models are generally small and within the typical error bars (∼0.15 dex), in general agreement with previous studies. The Al  iii to Al  ii ratio for these and other absorbers from the literature are compared, and we find that while the sub-DLAs have a larger scatter in the Al  iii to Al  ii ratios than the DLAs, there appears to be little correlation between the ratio and   N _{H  i}   . The relationship between the metallicity and the velocity width of the profile for these systems is investigated. We show that the sub-DLAs that have been observed to date follow a similar trend as DLA absorbers, with the more metal rich systems exhibiting large velocity widths. We also find that the systems at the upper edge of this relationship with high metallicities and large velocity widths are more likely to be sub-DLAs than DLA absorbers, perhaps implying that the sub-DLA absorbers are more representative of massive galaxies.     

Spectroscopy of SN 1987A at 0.9–2.4 μm: days 1348–3158

We present near-infrared spectroscopic observations of SN 1987A covering the period 1358 to 3158 d post explosion. This is the first time that IR spectra of a supernova have been obtained to such late epochs. The spectra comprise emission from both the ejecta and the bright, ring-shaped circumstellar medium (CSM). The most prominent CSM emission lines are recombination lines of H  i and He  i , and forbidden lines of [S  iii ] and [Fe  ii ]. The ejecta spectra include allowed lines of H  i , He  i and Na  i and forbidden lines of [Si  i ], [Fe  i ], [Fe  ii ] and possibly [S  i ]. The intensity ratios and widths of the H  i ejecta lines are consistent with a low-temperature Case B recombination spectrum arising from non-thermal ionization/excitation in an extended, adiabatically-cooled H envelope, as predicted by several authors. The slow decline of the ejecta forbidden lines, especially those of [Si  i ], indicates that pure non-thermal excitation was taking place, driven increasingly by the decay of ⁴⁴Ti. The ejecta iron exhibits particularly high velocities  (4000–4500 km s^-1)  , supporting scenarios where fast radioactive nickel is created and ejected just after the core bounce. In addition, the ejecta lines continue to exhibit blueshifts with values ∼−200 to −800 km s⁻¹ to at least day 2000. These blueshifts, which first appeared around day 600, probably indicate that very dense concentrations of dust persist in the ejecta, although an alternative explanation of asymmetry in the excitation conditions is not ruled out.     

Helium recombination spectra as temperature diagnostics for planetary nebulae

Electron temperatures derived from the He  i recombination line ratios, designated T _e(He  i ), are presented for 48 planetary nebulae (PNe). We study the effect that temperature fluctuations inside nebulae have on the T _e(He  i ) value. We show that a comparison between T _e(He  i ) and the electron temperature derived from the Balmer jump of the H  i recombination spectrum, designated T _e(H  i ), provides an opportunity to discriminate between the paradigms of a chemically homogeneous plasma with temperature and density variations, and a two-abundance nebular model with hydrogen-deficient material embedded in diffuse gas of a 'normal' chemical composition (i.e. ∼solar), as the possible causes of the dichotomy between the abundances that are deduced from collisionally excited lines and those deduced from recombination lines. We find that T _e(He  i ) values are significantly lower than T _e(H  i ) values, with an average difference of  〈 T _e(H  i ) − T _e(He  i )〉= 4000 K  . The result is consistent with the expectation of the two-abundance nebular model but is opposite to the prediction of the scenarios of temperature fluctuations and/or density inhomogeneities. From the observed difference between T _e(He  i ) and T _e(H  i ), we estimate that the filling factor of hydrogen-deficient components has a typical value of 10⁻⁴. In spite of its small mass, the existence of hydrogen-deficient inclusions may potentially have a profound effect in enhancing the intensities of He  i recombination lines and thereby lead to apparently overestimated helium abundances for PNe.