© The Authors 2024

1 Introduction

NGC 4361 (PN G294.1 +43.6) is one of the rare planetary nebulae (PNe) with He II 4686 Å emission stronger than Hβ and its nebula ionisation class, defined by He II/Hβ (Dopita & Meatheringham 1990), exceeds 10 (the maximum value). Even given the temperature of hot PN central stars, such a condition is suggestive of a nebula optically thin in the Lyman continuum (i.e. matter-bounded). The high ionisation of NGC 4361 has been confirmed by the presence of other high-ionisation species from optical (Heap et al. 1969; Aller 1978; Torres-Peimbert et al. 1990) and ultra-violet (UV) (Adam & Koeppen 1985) spectroscopy. Aller (1978) reported a detection of [Ne V]3426 Å, indicating ionisation by >120 eV photons and Torres-Peimbert et al. (1990) from the [O III]4363/5007 Å ratio determined a high value of T_e of ~18 000 K, later confirmed by Liu (1998). The nebula is also detected in extended X-ray emission by EXOSAT (Apparao & Tarafdar 1989) and ROSAT (Kreysing et al. 1992), indicating a plasma at <2×10⁵K. Aller (1978) noted the extreme weakness of the low-ionisation lines and no studies have detected [N II] emission in NGC 4361 to date.

The bright central star (Gaia mean G mag. 13.0876) was observed by Mendez et al. (1981, 1992) and fit by a model atmosphere ~80 kK; however, the interpretation is complicated by the strong He II emission filling in the stellar absorption lines. Ziegler et al. (2012) analysed Far Ultraviolet Spectroscopic Explorer (FUSE) spectra and found a 126 kK central star (CS) with log g value of 6.0. Several attempts at modelling the nebular spectrum have been made: the early model of Aller et al. (1979) inferred typical Galactic PN metallicity 12 + log₁₀ (O/H) = 8.6 ([O/H]) and a CS of ~80kK, while Torres-Peimbert et al. (1990) suggested a low metallicity of [O/H] = 8.2 with a 90kK CS. Howard et al. (1997) found [O/H] of 8.15 fora 120 kK star. The model of Torres-Peimbert et al. (1990) showed an overall C/O ratio >1, but with indications of the inner zone being richer in C. While the high C/O was confirmed by the spectroscopy of Liu (1998) from an analysis of the C III and C IV recombination lines, the abundance of C was not confirmed to vary spatially. The electron density has been measured at ~1200–1500 cm⁻³ (Aller et al. 1979; Liu 1998). Given the Galactic latitude and possible evidence for low (O/H), Torres-Peimbert et al. (1990) suggested NGC 4361 as a Galactic halo PN, so of Population II low-mass progenitor, but given the distance from Gaia (π = 0.9653 ± 0.0439 mas, inferred distance 1040 pc), it may be in the thick disc.

The overall morphology is elliptical with hook-shaped extensions to the north-east (NE) and south-east (SW) without a clear differentiation of a shell. The morphology has been variously reported as filamentary and with a halo, but integral field spectroscopy by Monreal-Ibero et al. (2006) with the VLT Imaging and Multi-Object Spectrometer (VIMOS) found no true halo below ~10⁻² of the central surface brightness beyond a nebula extension of ~110″. From the kinematics, NGC 4361 was initially suggested to be bipolar based on long slit observations (Vázquez et al. 1999). The more extensive spatial coverage by Muthu & Anandarao (2001) suggested a double bipolar (quadrupolar) since velocity-split line profiles occur along NE- SW and NW-SE axes. The velocity profile separation of upto 70 km s⁻¹ (Vázquez et al. 1999) is notably high for a PN.

NGC 4361 was included in the MUSE commissioning on account of its high emission line surface brightness and its size, which overfills the MUSE field of view (60×60″). Other PNe observed in MUSE commissioning are NGC 3132 (Monreal-Ibero & Walsh 2020), IC 418 (Monreal-Ibero & Walsh 2022) and IC 4406 (Ramos-Larios et al. 2022). With their depth, field coverage, spatial resolution, and spectra over the optical range, these MUSE observations bring important new data to bear on the nature of this extreme PN, in particular detecting low-ionisation emission for the first time, but with a surprising structure. Section 2 describes the MUSE observations and the major reduction steps to the data cubes. Section 3 presents and describes the emission line and line ratio maps and Sect. 4 gives the extinction and line diagnostic images. Sect. 5 presents analysis of the many compact [N II] knots, whose striking contrast to the appearance of the nebula in all lines of higher ionisation and their small size, led us to nickname NGC 4361 the ‘freckles’ nebula. Section 6 then presents a discussion of these freckles and their relation to the high-ionisation nebula. In the outer field, a serendipitously discovered galaxy ‘shining’ through the nebula was identified (Bacon et al. 2014) and in Sect. 7 an analysis of this target is presented. Section 8 draws together the conclusions.

2 Observations and reductions

NGC 4361 was observed during the second MUSE commissioning run in May 2014 over five nights. MUSE (Bacon et al. 2014) at the Very Large Telescope (VLT) was deployed in standard Wide Field Mode (WFM, 60×60″ field with 0.20″ spaxels) with the default wavelength range 4750–9300 Å at 1.25 Å sampling (MUSE WFM-NOAO-N mode). A total of 74 exposures of 60 s each were obtained, with 58 centred on the central star of the nebula and 16 at an offset position 30″ W to sample the extent of the nebula. Figure 1 shows the central pointing overlaid on the Spitzer Infrared Array Camera (IRAC) 3.5 + 4.5 µm image. Between exposures 90° rotations (PAs of 0, 90, 180, and 270° were covered) and dithers of 0.4″ offsets were generally applied in order to average out the pattern of the 24 individual integral field units of MUSE. On account of poorer seeing and low transparency during some exposures, a selection was made based on measured flux in a narrow passband centred on Hα (the passband of the Hubble Space Telescope Wide Field Camera 3 ultraviolet and visible arm F656N filter was used, half power points 6552–6572 Å) during the reduction chain, leading to 16 (all from 2014-05-03) and three exposures being dropped from the centre and offset field, respectively. Table 1 summarises those observations that were selected. The differential image motion monitor (DIMM) seeing (where available) is listed in column 7.

The MUSE observations of NGC 4361 were reduced with EsoRex (version 3.13) scripts and the MUSE instrument pipeline (version 2.8) (Weilbacher et al. 2014, 2020). Since these were commissioning observations, the full set of calibration files was not necessarily available for each night. Thus, sets of compatible bias frame, master flat, and wavelength calibration exposures were assembled from the closest available calibrations to produce master bias, master flat, trace table, and wavecal tables appropriate for each night. For example, wavelength calibrations for the exposures on 2014-05-07 were not available and those for 2014-05-06 were employed. The mean spectral resolving power from the fit of the muse_wavecal task to the arc lamp exposures was 3020±75. The line spread function profile fits required for the extraction of the spectra were also produced using calibration files generated for each night (or for the closest possible date when not available). A suitable sky flatfield was available for 2014-05-06 and was used to produce the necessary skyflat for all datasets. The bad pixel and geometry tables developed for commissioning (calib-0.18.2) were used in the (muse_scibasic) extraction of the spectra for formation of the pixel tables.

Observations of the white dwarf standard star GD 108 (Bohlin et al. 2014; Oke 1990) from 2014-05-06 were reduced similarly to the NGC 4361 observations and a single response table produced, which was applied to the data for all five nights. Since it was not necessarily known from the nightlogs and the DIMM results which exposures were affected by poorer visibility and/or light cloud, an Hα narrowband (6552–6572 Å) image was produced for each exposure and the total flux used to flag the exposures suffering poorer visibility. Using exposures in a V′ band filter (square profile, 5350–5650 Å) and centroiding (IRAF¹ imexam) the central star, the offsets between exposures were determined and used in aligning the individual exposures to produce a single combined data cube. For the W offset position, the central star was at the extreme edge of the field and could not be reliably used for image alignment; since no other stars were bright enough for alignment purposes, the default telescope pointing enscribed in the headers was used by default to align the multiple exposures. As a result, the image quality and alignment is not as good in the offset field as the central pointing. For the central field, 42 observations over three nights were combined (see Table 1), and the resulting V′-band (5350-5650 Å) image quality measured in the final combination was 0.74″ (Gaussian FWHM of the central star). For the offset field, the resulting image quality for the combined exposures was somewhat worse (∽1.0″); however, from the signal in the overlap region of the central and offset fields, the W offset field was found to have only 17% of the depth of the central field (lower exposure time and clouds).

For the central field, sky subtraction proved to be problematic as there is little area on the MUSE field occupied by sky uncontaminated by the extended nebulosity. For each position angle, a careful assessment of the minimum area to use for sky was made and two small triangles to the SE and NW, consisting of a total area of ∽1900 spaxels (76 arcsec²), 1.85% of the MUSE field was chosen (the extents of these two regions are indicated by the dark blue shaded areas on the Hβ image, Fig. 2). For the outer field, nebulosity only extends westward over about two thirds of the MUSE field, and 8% of the field was adopted as sky. Some weak nebula emission nevertheless occurs over these adopted sky regions for the central field, so that the total flux of the extended line emission (e.g. H, He) in the MUSE field will not be absolute. However, since the MUSE field does not cover the whole nebula extent, and absolute fluxes are not sought, this effect can be neglected. An indication of the accuracy of the sky subtraction shows that the extended [N I]5199,5202 Å sky line was no longer detectable in the total spectrum. For analysis of the spatially compact ‘freckles’ (Sect. 5), small errors in sky subtraction are uncritical.

The resultant sky-subtracted cubes have dimensions 426 (α) by 433 (δ) [61.2 × 61.3″] by 3640 (λ) voxels (4750–9300 Å) at the default binning of 1.25 Å. The absolute astrometry was adjusted using coordinates of the stars in the Gaia DR3 catalogue (Lindegren et al. 2021) which are available in the MUSE field of view.

Fig. 1 Location of the central MUSE field is shown, superposed on an image of the wider field including the nebula extent, from the Spitzer IRAC combined 3.6 and 4.5 µm image. The scale and orientation are indicated. Image credit: NASA/JPL-Caltech.

Fig. 2 Images of NGC 4361 in Hβ, He I 5876 Å and He II 5412 Å with linear colour table lookup (see the colour bar right of each figure). The two coloured wedges at the SE and NW corners of the field shown on the Hβ image represent the regions used for sky background estimation. The contour map on the images is derived from the Hβ image.

3 Results: Spectral imaging

The cubes were analysed with a semi-automatic Gaussian line fitter task already described in Walsh et al. (2018), Sect. 3.2. The line list to drive the fitting for the spectrum in each spaxel was derived from the list of emission lines identified by Aller et al. (1979), supplemented with a manual identification of the emission lines present in an integrated spectrum of the nebula over an area 60″ × 60″ centred on the CS (see Table 2). The predominance of high-ionisation lines and the extreme weakness of low-ionisation lines (as found in earlier studies) is confirmed. However, the depth of the spectra (amounting to 42 × 60 s = 2520 s per spaxel) conclusively lead to the detection of low-ionisation lines of [N II], [O II], and [O I] for the first time.

A selection of example spectra of NGC 4361 are shown in Fig. 3: the full (4750–9300 Å) spectrum, on a log flux scale, of a 1 arcsec² region over the bright shell in the core (Δα − 5.2″, Δδ + 6.4″ with respect to the position of the CS), along with three red spectra (6500–6750 Å) of freckles (see Sect. 5 for details). Freckle 9 is one of the fainter ones in terms of Hα and N II 6583 Å flux (lower-left); Freckle 12 is of average brightness (centre-bottom) and Freckle 94 is the brightest (lower-right). Section 5 provides a detailed description of the freckles, their position, and numbering.

From the Gaussian fitted line flux per spaxel, the emission flux and error maps were derived. Over the image of the CS, the continuum is strong and displays absorption and emission lines, so subtraction of the stellar spectrum is necessary to reveal the nebular emission morphology in this area. The spectrum of the CS was determined by extracting the star+nebula over an area of 530 spaxels (radius 2.6") and then subtracting the nebular emission from an annulus over radii 3.0-3.8". The resulting stellar spectrum (presented in Appendix A) was then subtracted from each spaxel spectrum over the central area (5×5″) by scaling the stellar continuum over five windows (4770–4830, 5130–5370, 5450–5780, 6800–7000, and 7350–7510 Å), all selected as without strong nebula emission lines (however, still containing a contribution of some weak lines). This effectively removes the stellar contribution over the central area before Gaussians were fitted to the emission lines, but neglects the very minor nebular continuum contribution. A selection of these flux images is shown in Figs. 2, 4, and 5. The (rest air) wavelengths and integrated fluxes of the selected lines displayed are listed in Table 3 for reference.

Figure 2 show the images of the Hβ (4681.3 Å), He I 5876.0 Å, and He II 5411.5 Å (recombination) line emission. The contour map shown overlaid on the line images is derived from Hβ. Figure 4 contrasts the morphology between the lines of the same element for differing ionisation level, [Ar III]7135.8 Å (ionisation potential, I.P., 40.7 eV) and [Ar V]7005.7 Å (I.P. 74.8 eV), and also shows the bright medium ionisation level [O III]5006.8 Å (I.P. 54.9 eV) and [S III]9068.6 Å (I.P. 34.9 eV) line images.

Figure 5 shows the striking morphology of the low-ionisation emission as exemplified by the strongest low-ionisation line, [N II]6583.4 Å. In contrast to higher ionisation species, the emission is mostly prominent as compact knots, some in striking radial orientations. These were dubbed ‘freckles’ on account of their unexpected appearance and frequency over the face of the nebula, in comparison to the other emission line images. From the [N II]6583 Å image, a total of 102 compact knots were counted by eye. There is in addition very faint extended [N II] emission similarly following the morphology of the [O III] line for example (brighter regions to WNW and E). More details on the ‘freckles’ are provided in Sect. 5, where an analysis of the extracted spectra is presented.

The morphology is remarkably similar between Hβ and He II 5412 Å as shown in Fig. 6. The flat appearance of He II/Hβ (ratio 0.089 ± 0.0073) continues into the outer field (not shown on account of a lower signal-to-noise ratio, S/N) without an obvious decrease to lower values, reinforcing evidence of an optically thin nebula. When plotting the ratio He I 5876 Å/He II 5412 Å, it is clear that He⁺⁺ is beginning to recombine to He⁺ towards the edges of the field, although the ratio is low (0.095 ± 0.044); several of the low-ionisation freckles appear in this plot since they contain singly ionised He. That recombination occurs to the outer edges of the MUSE field is evident from the [O III]5007 Å/Hβ ratio image (Fig. 6), where higher ionisation species of O (O⁴⁺ and O³⁺) in the central region must be recombining to O²⁺ outwards, predominantly to the NE, NW, and SW. The column of low [O III]/Hβ (NNW to SSE) is notable. Additional evidence for lower ionisation is found in the [N II] image, where faint extended emission is found at the peaks of [O III] emission to the E and NW. The ratio [Ar V]7006 Å/Hβ image is quite distinct from any other (Fig. 6) and shows a prominent elliptical ring (minor axis radius 19″) with openings at the ends of the major axis to ENE and WSW, aligned with the large-scale nebula extension seen in Fig. 1.

Fig. 3 Selection of representative spectra (log10 flux) of a few given regions. Top: full observed spectrum over a bright region in the core of the nebula (area 1 arcsec2 − 25 MUSE spaxels) centred at (Δα − 5.2″, Δδ + 6.4″) relative to the position of the CS. Bottom: three example spectra over freckles chosen to sample faint, average, and bright freckles (based on their extracted Hα and [N II] fluxes; see Sect. 5.2). Left: spectrum of a faint freckle (9). Centre: Spectrum of a freckle (12) whose Hα and N II fluxes mark it as of average brightness; spectrum of the brightest freckle (94). Note: these spectra of freckles include the (high-ionisation) background nebular spectrum within the area of the freckle.

Fig. 4 Images of NGC 4361 in two Ar ions – [Ar III]7136 Å and [Ar V]7006 Å – demonstrating the differing morphology at two ionisation levels for the same element (top). Images in the two strongest lines of medium ionisation species – [O III]5007 Å and [S III]9069 Å (bottom). The contour lines are derived from the Hβ image in Fig. 2.

4 Extinction and diagnostics of physical conditions

4.1 Extinction determination

From images of the ratios of Balmer and Paschen line fluxes compared to the Menzel & Baker (1937) Case B values, a map of the extinction across the nebula can be constructed. The value of the electron temperature, T_e , and electron density, N_e , must be provided, such as from collisionally excited line (CEL) indicators; for example, [S III] 6312/9069 Å and [Cl III] 5517/5538 Å, respectively. Since the He⁺⁺ lines are strong, then the contamination of Balmer and Paschen line fluxes by nearby (within ∼3 Å) He II emission must also be taken into account. The S/N per spaxel in the CEL diagnostic lines was however insufficient to compute the spatial distribution of T_e and N_e , so integrated values for the whole MUSE field were adopted, T_e = 17 000 K and N_e=1500 cm⁻³, see the following sub-section.

Figure 7 shows the resulting extinction map of c, the log extinction at Hβ, from the ratio of Hα/Hβ, correcting Hβ for the presence of He II 4859.3 Å, n=4–8 and Hα for the presence of He II 6560.10 Å, n=4–6. The fluxes of the He⁺⁺ lines are given by the flux of the He II 5412 Å (n=4–7) line and the ratios of the He II Pickering series from the computations of Hummer & Storey (1987), adopting Case B, T_e=17 000 K and N_e=1500 cm⁻³ . The mean value of c in Fig. 7 is 0.096 ± 0.031 with three rounds of 3σ rejection (mean S/N of 4.7 per spaxel); the mean value is higher by 0.009 if the corrections for the two He II lines are neglected. If Case A for He II is adopted, then the mean value of c is higher by 0.001. Including the Paschen lines into the determination of the c map, has almost negligible effect, and increases the mean c value by 0.001, while increasing the error on account of the much lower Paschen line fluxes. Figure 7 shows the resulting extinction map, with lower values over the central 30″, increasing slightly to the NE and SW. Over the area of the central star the strong absorption lines of H I and He II hinder a clean determination of the H Balmer and Paschen line fluxes and the He II 5412 Å line flux, and so the value of c over this area was substituted by the mean of the surroundings. The observed emission line flux images were all dereddened using the c map shown in Fig. 7 where He II line contamination of the Balmer lines was taken into account.

Fig. 5 Strikingly contrasted appearance of NGC 4361 at low ionisation is shown by this image of the brightest optical line [N II] 6583 Å.

4.2 T_e and N_e determination

4.2.1 CEL diagnostics

The collisionally excited lines (CELs) in NGC 4361 are all weak with respect to H and He, and therefore the available CEL diagnostics for electron temperature and density ([N II], [S III] and [Ar III] for T_e, [S II] and [Cl III] for N_e) have very low S/N per spaxel, effectively precluding making images of T_e and N_e at the original spaxel resolution (0.2″). The summation of spaxels into successively larger bins was attempted until sufficient a S/N was reached to achieve a fully filled image of T_e or N_e . For example, this was achieved with T_e from [S III] 6312/9069 Å with 5×5 spaxel bins (1.0″). However for T_e from [N II] 5755/6583 Å the bins had to be at least 20×20 (4.0″) before an image of sufficient S/N could be produced to compute an image filled with valid T_e values. For N_e from [S II] 6716/6731 Å and [Cl III] 5517/5537 Å the lines are even weaker and even 4.0″ bins were not adequate. Thus, only for T_e from [S III] could an adequately sampled image be produced; for the other diagnostics a single value was computed based on the integrated spectrum of the whole field. All determinations of H and He and metal CEL line ratios were made with PyNeb (Luridiana et al. 2015). The sources for the atomic data employed for all the PyNeb CEL computations are listed in Appendix B.

However for [S III] T_e, the range of many of the values of the ratio 6312/9069 Å (dereddened line fluxes) was beyond the feasible range of the ratio of 0.0066–0.2096 (for N_e ≤ 2000 cm⁻³). Various factors could cause this condition, such as telluric absorption of the red [S III] 9068.6 Å or a contaminating line at or close to the wavelength of the weaker [S III] 6312.1 Å. Indeed there is an He II n = 5 series line at rest wavelength 6310.85 (5–16), close enough to 6312.1 Å to be included in the integrated flux of the [S III] line at the MUSE spectral resolution of ~2.5 Å. From the dereddened flux of He II 5412 Å and the Case B line ratios from Hummer & Storey (1987) for an adopted T_e of 17 000 K and N_e of 1500 cm⁻³, the ratio 6311/5412 Å = 0.04623. Subtracting the expected flux of this line lowers the flux of the fitted 6312 Å flux by ~40%. With this correction the 6312/9069 Å ratio enters feasible values and T_e ([S III]), can be calculated. Figure 8 shows the resulting image, where the mean T_e is 14 940 ± 3800 K, but a very noticeable increase to the eastern lobe of the nebula (cf., the gradient in [O III] T_e shown by Liu (1998), Fig. 4a for a NS slit, positioned 10″ E of the central star).

Table 2 presents the observed line fluxes (normalised to I(Hβ)=100), calculated by Gaussian fits to the detected emission lines (as in Sect. 3), for the total spectrum within the MUSE central field (60″×60″). Column 4 presents the error on the line flux from the Gaussian fit (with the error on Hβ set to zero and so propagated to all other flux errors). The extinction determined from Hα/Hβ, with correction for the presence of He II lines, is also listed in Table 2. Column 5 lists the dereddened line fluxes and Column 6 the propagated errors. Table 4 then presents the CEL diagnostics for the whole MUSE field based on the dereddened fluxes listed in Table 2. The N_e and T_e error values were determined by Monte Carlo based on the (1σ) flux ratio errors for 1000 trials. The value for T_e ([S III]) is included for comparison with the value from the binned spaxel determination (Fig. 8). On account of the high T_e values and the large errors on the density sensitive [S II] and [Cl III] line ratios, in PyNeb it was found to be more reliable to compute the diagnostics from the single line ratio, rather than the getT emDen task, using the appropriate T_e or N_e values, after iteration. For the integrated spectrum of the full MUSE field, the [Ar III] 5191.8 Å is detectable, so the determination of T_e from the 5192/7136 Å becomes feasible and the value is listed in Table 4.

Fig. 6 Ratio images of NGC 4361 in He lines (top): He II 5412 Å/Hβ (left) and He I 5876 Å/He II 5412 Å (right) demonstrating that a very large fraction of He is fully ionised throughout the nebula. Ratio images displaying the ionisation zones (bottom): [O III] 5007 Å/Hβ (left) for the medium ionisation zone and [Ar V] 7006 Å/Hβ for the highest ionisation (right).

4.2.2 ORL diagnostics

The ratios of H⁺ , He⁺, and He⁺⁺ lines can also be used for T_e, N_e determination. In particular, the ratios of He I singlet lines can be used for T_e determination (Zhang et al. 2005; Walsh et al. 2018); the ratio of higher H Paschen lines can be used for N_e measurement (Zhang et al. 2004; Walsh et al. 2018); and the H I Paschen jump (8204 Å) is sensitive to T_e (Zhang et al. 2004; Walsh et al. 2018), as is the He II n=5 series limit jump at 5694 Å (Fang & Liu 2011). Given the extreme weakness of He I lines (Tables 2 and 3 and Fig. 2), no attempt was made to use the line ratios for T_e estimation.

Employing the Hummer & Storey (1987) Case B emissivities for He II, an attempt was made to use the ratios of brightest n=5 lines (n=5–9 8236.8 Å, n=5–10 7592.8 Å, n=5–11 7177.5 Å, n=5–12 6890.9 Å, n=5–13 6683.2 Å, n=5–14 6527.1 Å, n=5– 15 6406.4 Å, and n=5–17 6233.8 Å) to n=4–7 5411.5 Å) to estimate T_e ; however there is too little dependence of the ratios of these lines with T_e at the available S/N and accuracy of the flux calibration, even for integrated spectra, to derive any likely values within 5000–35 000 K. Given the evidence that the nebula is optically thin, Case A emissivities are suggested, but the same conclusion results as for Case B.

N_e (H I) determination from higher Paschen line ratios (P15– P26, [8545.4, 8502.5, 8467.3, 8438.0, 8413.3, 8392.4, 8374.8, 8359.0, 8345.5, 8333.8, 8323.4, and 8314.3 Å), cf. Walsh et al. (2018), was also attempted. However the quality of the flux calibration and the proximity of some lines to nearby strong He II lines of the n=6 series, affects the line fitting of the H Paschen lines, and overall did not allow a useful estimate of N_e to be determined without large uncertainties.

The H I Paschen jump (PJ) at 8204 Å is prominent in the spectra across the MUSE field and the Paschen jump was calculated in the same way as in Walsh et al. (2018). However, on account of the strength of the He II lines, the regions selected to be free of all but very weak emission lines for measurement of the continuum on both sides of the jump were altered from those given in the Appendix A of Walsh et al. (2018). To the blue of the jump, the region 8100–8180 Å was chosen and to the red of the jump, regions 9115–9140 and 9170–9215 Å. An attempt to form an image of the jump, normalised by the Paschen 11 (8862.8 Å) emission line flux, (called PJ/P11), required binning of the spax- els to reach an image not dominated by large value fluctuations, at the level of at least 10×10 spaxels (2.0″) but showing little structure not related to the flux level of the nebular continuum (cf., Hβ image, Fig. 2). However the region east of the CS, which displays an elevated value for T_e ([S III]) (see Fig. 8), showed slightly higher PJ/P11 (lower PJ T_e), while the region west of the CS with lower T_e ([S III]) had a lower PJ/P11 (higher T_e). However, these PJ/P11 values suggested very high temperatures (>25 000 K), close to the cut-off value of 30 000K (the Hummer & Storey (1987) H line emissivity tabulations do not go beyond this value) and the PJ T_e errors (estimated by Monte Carlo based on 100 trials with the 1σ flux errors delivered by the MUSE pipeline) are of similar magnitude to the T_e values.

Therefore, given the generally low S/N of the Paschen jump even in binned spectra, the Paschen jump was computed only in large integrated regions: the full MUSE field, with a region of radius 1.8″ around the central star excluded; the region of high T_e ([S III]) (650 arcsec²) in Fig. 8; the region of lower T_e (702 arcsec²) in Fig. 8. Since the freckles are distinct regions and (as shown in the following section) of lower ionisation and electron temperature than the bulk nebula, a separate region of the whole field, excluding the central star and the area of the emission of all the freckles (189 arcsec²) was formed. Table 5 lists the PJ/P11 values determined from the continuum sections defined. Column 4 lists the derived Paschen jump T_PJ values, using the same methodology described in Walsh et al. (2018) to compare the observed PJ/P11 to the theoretical values with the modified blue and red continuum extents. A value of 0.099 was calculated for the He/H abundance (He⁺/H⁺ = 0.0041, He⁺⁺/H⁺ = 0.0951) based on the strengths of the He I 5876 Å and He II 5412 Å lines for a single value of T_e of 17 000K and N_e of 1500 cm⁻³.

However, when a graphical comparison of the dereddened spectra of the regions listed in Table 5 with the computed nebular continuum (including H I bound-free (bf) transitions, free-free and 2-photon (2ν) , He I bf & ff, and He II bf, ff, & 2ν) was made, the computed He II n=5 continuum jump at 5694 Å was too large with respect to the local continuum in this vicinity. The implication is that the T_e (He II) is larger than for H I. This was not unforeseen given previous determinations of the He II jump T_e, cf. Liu & Danziger (1993), Fang & Liu (2011) which showed elevated T_e (He II) compared to T_e (H I) from PJ/P11. Given the high abundance of He⁺⁺ , amounting to 95% of the He abundance, then the contribution of the He II (n=6) jump to the H I Paschen jump is significant (18%) so that the assignment of the T_e (He II) plays an important role in determining Te (H I) from PJ/P11. The values of T_e (H I) from PJ and He II from n=5 jump were thus calculated iteratively. From the value of PJ T_e listed in Table 5, the H I and He I continua were computed and subtracted from the dereddened spectrum of the NGC 4361 integrated regions, leaving the He II continuum to be included². The He II jump at 5694 Å was then fitted matching the jump graphically (an approach using continuum bands to the blue and red of the jump as for the Paschen jump was not successful on account of the difficulty of finding line-free continuum regions in this wavelength range). This value was then used to compute the Paschen jump allowing the T_e (H I) to be a variable. The resulting H I PJ T_e was then lower than the initial value (assuming a single value of T_e for H⁺ and He⁺⁺) on account of the higher T_e (He II), which lowered the He⁺⁺ contribution to the Paschen jump. Table 5 lists the resulting H I and T_e (He II) values. No error evaluation as such was performed but typical errors of ±1000 K are indicated for the T_e (H I) and ±2000 K for He II on account of the much weaker break.

Fig. 7 Log extinction at Hβ, c, image of NGC 4361 is shown for the central MUSE field. The area over the bright central star has been substituted by a constant value (see text for more detail). The contour lines are for Hβ flux as in Fig. 2.

Fig. 8 Te image from [S III] 6312/9069 Å. The original MUSE spax- els were binned 5×5 (1.0×1.0″) to increase the S/N in the fainter line (6312 Å).

5 Census of the NGC 4361 freckles

5.1 Identification and location

The freckles are the most obvious features on the [N II] image on account of their compact emission superposed on a very low background (Fig. 5). However in the strong emission lines of H and higher ionisation CELs, the features are difficult to distinguish. The brightest freckle (94, Δa −2″, Δδ +27″, see below) has a contrast in the Hα image of ∼40% with respect to the surroundings, the 2nd brightest freckle a contrast of 14%, and the rest of the freckles having an Hα contrast <10%. 102 separate freckles were identified on the [N II] image by eye and Table 6 presents their designation, centroid positions as Δα, Δδ with respect to the position of the central star (Gaia DR3 coordinate: 12^h 24^m 30.75^s, −18^deg 47′ 5.73″), their area as delineated (to ∼3σ above the background emission) in arcsec² and the [N II] 6583 Å flux (but including the background contribution). The faintest freckle is 0.3% of the brightest. By excluding the area of all the freckles, an integrated spectrum of the high-ionisation nebular component can be formed, as referenced in Table 5. The designation of the freckles is shown on the [N II] 6583 Å image (Fig. 10).

Column 6 of Table 6 lists the [N II] velocity and Column 7 the difference in velocity between the [N II] and Hα emission (fitted by single Gaussians) over the extent of each freckle. On account of the relative strength of the Hα and [N II] emission over the freckles compared to the background, the [N II] velocity measured that of the freckle, while the Hα velocity that of the local bulk nebular emission. At the MUSE spectral resolution (∼130 km s⁻¹ at Hα), it was not feasible to deconvolve the Hα velocities of the freckle and the local bulk ionised gas. The minimum and maximum [N II] − Hα velocity differences are −34 and +45 km s⁻¹, and the mean +3.4 km s⁻¹, to be compared to the systemic (heliocentric) velocity of the nebula at Hα of +5 km s⁻¹ for the full MUSE central field. The typical error on a single velocity determination is at least 10 km s⁻¹ for a high line flux, so there is no evidence that the system of freckles has a bulk displacement from that of the overall nebula. Figure 10 displays the large-scale Hα velocity field for the central MUSE field and the numbering of the freckles; the size of the freckle marker is proportional to the [N II] flux and the offset velocities of the freckles from the bground emission (Table 6, Column 7) are colour-coded.

An examination of the spatial distribution of [N II] freckles shows that the SW quadrant has the lowest number whilst the other three quadrants are fairly similarly populated. There are some obvious close groups of freckles, such as 31–37, 38–41 and 54–60. Examination of Table 6 shows that 31–35 and 37 have similar +ve velocities with respect to that of the surroundings, 38-41 also with low velocities, and 55–59 have similar −ve velocities. It is clear here (and from correlations of the positions with the velocities in Table 6) that freckles projected close on the sky may not share the same velocity – and presumably their line-of sight position (e.g. assuming that a −ve velocity implies a freckle foreground to the local emission). There are also a few obvious linear alignments: to the SSE, 2, 6, 8, 11, 16, and 17 (all with −ve velocities in range −17 to −26 km s⁻¹); to the NNW, 80, 96, and 97; (+ve velocities in range 20–36 km s⁻¹); and to the NW, 82, 83, 84, 86, and 91 (velocities in range –9 to +5 km s⁻¹). The SSE alignment points back to the central star within about 1″, as does the NNW one; however, the NW one does not and is tangent to the SW of the CS by 8″. None of these alignments show convincing evidence of a coherent velocity gradient, as might be indicative of an ejection or acceleration sequence, although the MUSE velocity resolution is rather low.

Fig. 9 Dereddened spectrum of the full MUSE field of NGC 4361, with the area of the central star subtracted, is shown with the fit of the H+, He+ and He++ nebular continua (in blue, yellow and red, respectively). The sum of the three sets of continua is shown in green. The fit to the spectrum is shown in magenta and includes a scattered light contribution modelled as a third-order polynomial continuum to match the sum of the nebular continua to the measured nebular spectrum. See text for further details and Table 5 (Columns 5 and 6) for the fitted jump temperatures (H I Paschen jump and He II n=5 jump).

Fig. 10 Velocity field for the central MUSE field NGC 4361 determined from Hα is shown together with the numbering for the freckles, with a circle for each proportional in size to the [N II]6583 Å flux. The Hα and individual freckle offset velocities ([N II] − local Hα), in km s−1, are colour-coded as shown by the colour bar on the right.

5.2 Spectra

The properties of the freckles, as distinct from the bulk emission of the nebula, such as their ionisation, extinction, and T_e, N_e diagnostics can be revealed by emission line spectroscopy. However, as explained in Sect. 5.1, their observed properties as faint compact sources on a high and structured nebula background introduce severe difficulties in extracting their intrinsic spectra. A number of approaches were tried to reach the highest number of non-zero emission line measurements among the 102 freckles:

• For each wavelength slice in the MUSE cube, a 2D gaussian- smoothed surface was fitted to the whole field excluding the areas of the freckles, then the spectrum of the background subtracted emission was extracted for each freckle. This approach performed poorly on account of small scale structure in the background and non-uniformities in individual wavelength slices;

• For each wavelength slice in the cube, aperture photometry was performed with a circular aperture for the background (excluding spaxels occurring over any other freckles), subtracted from the signal over the extent of the freckle and the integrated spectrum formed. The approach also performed poorly and gradients in the background emission often compromised the estimation of a single background value over the freckle area;

• Given the velocity of the freckle from [N II] and the velocity of the local background from the strong Hα emission, the total emission line over the freckle was simulated by a sum of Gaussians at the background and freckle velocities, then fitted for the fluxes. Performance as judged by the quality of the fits to bright lines, such as Hα or [O III], was generally unsatisfactory on account of the low velocity resolution of MUSE, the generally small velocity separation of freckle and background and small deviations from true Gaussians for the line profiles;

• A simpler approach was to apply local background subtraction to the emission line maps already produced as described in Sect. 3 and shown in Figs. 2, 4, and 5. Rather than circular background apertures the shape and size of the aperture was tuned to the location of nearby freckles and the direction of slope of the background emission. As estimators, both the median value over the area of the background external to the freckle and a bilinear fit, resulting in removal of a sloping background, were tested. An inspection of the subtracted background, showed that the latter produced the better fidelity background removal, although the median value produced more cases of feasible Hα/Hβ and [O III]/Hβ ratios, simply because more background emission was included in the area of the freckle.

Table 7 presents the set of line fluxes for all the freckles (designations as in Table 6), determined from the last method (with bilinear background estimate). For all the freckles, fluxes for a set of lines (H I, He I, He II, [O I], [O II], [O III], [N II], [S II], and [S III]) are tabulated; but for multiple freckles, some or many of these lines clearly have large errors as indicated by infeasible values, or are undetected. The factors that affect the quality of these spectra range simply from low fluxes for fainter freckles (see Hα flux listed in Table 7, column 2), to the strength of the background emission (particularly affecting freckles closer to the CS), the proximity of neighbouring freckles (cf. the tight groups 39–42 and 57–61) and the local gradient and complexity of the background nebular emission. Whilst the detection of [N II] 6583 Å is simple (strong line on a weak background), fluxes for lines with strong background contribution, such as Hα, Hβ, and [O III] could be subject to large systematics, and, for example, line ratios (Hα/Hβ and [O III] 4959/5007 Å) were measured outside the allowed ranges. Therefore, we decided to rank the quality of the freckle spectra into three quality bands:

• A

  – with high Hα flux (>1.5×10⁻¹⁶ erg cm⁻² s⁻¹), most lines detected, 2.5 ≤ R([O III] 5007/4959 Å) ≤ 3.5, and Hα/β >2.8;

• B

  – with many lines detected and one of 2.0 ≤ R([O III] 5007/4959 Å) ≤ 4.0 and 2.4 ≤ Hα/β ≤ 6;

• C

  – neither the R([O III] 5007/4959 Å) nor Hα/Hβ within the bounds of band B.

For category A spectra, extinction can be determined by comparison of Hα/Hβ to the Case B value, and usually N_e from [S II] 6716/6731 Å and T_e from [N II] 5755/6583 Å, and [S III] 6312/9069 Å. For category B spectra extinction can also be determined, and often N_e from [S II] 6716/6731 Å and occasionally T_e from [N II] 5755/6583 Å, and [S III] 6312/9069 Å. For C spectra, extinction cannot be determined. The total number of spectra in each quality band were: A 11; B 31; C 58; these quality bands are listed in the last column of Table 7.

Only those lines that were detectable in the freckles are listed – notably lines with ionisation energy ≲45 eV (corresponding to detection of [O III] and [Ar III]). In many freckle spectra, there is weak He II 5412 Å detected; however only for freckle 94 is this line clearly detectable on the He II 5412 Å image (Fig. 2, right) and its dereddened flux is only 1.7 (I(Hα) = 100). The spectra have (unusually) been normalised to Hα, since it is ~3× brighter than Hβ for both background and freckle, so the contrast of freckle to background is greater and the errors are lower for Hα than Hβ, resulting in the detection of a larger number of freckles.

The extinction for each freckle was determined from the Hα/Hβ ratio compared to Case B for assumed values of T_e = 11 000K and N_e = 1500 cm⁻³ (mean of T_e ([N II]) and N_e ([S II]), see Table 8). Table 9 then presents the [N II] 6583 ÅHα ratio, and, for A and B quality spectra, the extinction, and those physical diagnostics, where the line ratios are available. Figure 11 shows histograms of the integrated (observed) Hα flux in the freckles, the [N II] 65 683 Å/Hα ratio, the extinction coefficient c, N_e ([S II]) from 6716/6731 Å ratio, the T_e ([N II]) from 5755/6583 Å and the T_e ([S III]) from 6311/9069 Å ratio. Since some of these diagnostics were not available for all freckles, the number included in the histogram is listed in the title. Shown on all the plots except for Hα flux, are the equivalent values for the integrated nebula (see Sect. 4).

A thorough error analysis of the freckle fluxes (Table 7) was not attempted, but the errors over the freckle areas and the subtraction of the local background (based on the flux errors resulting from the Gaussian fitting of the emission lines, Sect. 3) were propagated. For the brightest freckle (94) the errors range from <0.5% for [N II] 6583 Å (bright line on a low background) to 25% for the faint line H I P9 9229 Å. For Hβ, a bright line but on a high background, the propagated error was 2.5%. For a freckle whose flux is at the peak of the histogram (see Fig. 11), for instance, 71, Hα flux 1.4×10⁻¹⁶ erg cm⁻² s⁻¹, the errors range from 11% on [N II] 6583 Å to 106% for [Ar III] 7751 Å. For fainter lines these values are typically exceeded, depending on the position of the freckle in the nebula, the background nebular emission, and the close proximity of other freckles. However it must be considered that systematic errors arising from the fitting to the background around the freckle and its subtraction may be much larger. For example, even for the brightest freckle, the difference between the total flux resulting from median and bilinear fitted backgrounds is 8% for the [N II] 6583 Å line, indicating the level of uncertainty.

The total spectrum of all freckles is presented in Table 10, together with the totals for the A and A+B quality spectra. The He II 5412 Å line is listed for the total freckles spectrum, but may be a systematic error from inadequate subtraction of the weak freckle flux from the bright He II background emission (except for 94). The summed line fluxes can also be computed including those above some cut-off. Table 8 summarises the cut-off Hα flux used in computing the sum, the resulting number of freckles summed, the extinction from Case B (assuming T_e = 11 000 K and N_e = 1500 cm⁻³), N_e ([S II]), T_e ([N II]) and T_e ([S III]) (with the flux of [S III] 6312 Å corrected for contamination by He II 6311 Å). It is clear from this table that the extinction increases as the freckle Hα flux increases (and for [N II] 6583 Å flux, given that [N II] 6583 Å/Hα is almost independent of freckle brightness). Also the N_e ([S II]) increases with brightness, while the T_e ([N II]) seems to be independent of freckle brightness. The T_e ([S III]) shows indications of a decrease in T_e with brightness but given the very high T_e values where this line ratio is not sensitive, this result may be an effect of increasing S/N as few higher flux freckles are summed. The alternative suggestion to very high T_e is of very high N_e ; for example a value of Ne ≳ 2.7×10⁵ cm⁻³ is needed to be consistent with T_e ([S III]) of < 20 000 K. However there is no evidence for such high values from N_e ([S II]).

Fig. 11 Histograms of the properties of the freckles: the observed Hα flux (log erg cm−2 s−1) (top left); the [N II]6583 Å/Hα ratio (upper right); log extinction at Hβ, c from Hα/Hβ (centre left); Ne ([S II]) (cm−3 (centre right); Te ([N II]) K (bottom left); Te ([S III]) K (bottom right); The vertical bar in the plots shows the corresponding value for the large- scale nebular emission of NGC 4361 from the MUSE central field, but excluding the area of the freckles. Note: each histogram only plots the available number of values of each parameter (listed in the title to each plot as (#) ); for the extinction, Ne and Te histograms only the values from the A and B quality spectra are shown.

6 Discussion

6.1 Optically thin versus optically thick

There is abundant evidence that NGC 4361 is optically thin to Lyman continuum radiation, but with the detection of low- ionisation species (in particular, the ‘freckles’ that are strong in [N II]), the question is now raised whether the nebula is density bounded at its outer extremities. The qualitative evidence that a PN is optically thin typically derives from its lack of extended low-ionisation line emission, particularly in the outer emission zone. The quantitative evidence that a nebula is optically thin usually derives from a comparison of the H I and He II Zanstra temperatures. The Zanstra temperature (Zanstra 1931) is derived by comparing the ionising flux required to produce a given emission line flux (either an H I or an He II line), compared to the flux in the central star continuum at some measured wavelength (typically in the optical range close in wavelength to the emission line in question). The relation between the stellar flux measured in the optical and the far-UV ionising flux required to produce the line emission can be provided by either a black body (BB) assumption or by a model atmosphere. Several authors have determined H I and He II Zanstra temperatures for NGC 4361 – Phillips (2003) (41³ and 93kK, BB assumption), Mendez et al. (1992) (43 and 99kK, nonlocal thermodynamic equilibrium (NLTE) model atmosphere). A more fundamental approach to determining the optical depth of a nebula is to construct a photoionisation model that matches the emission and general appearance of the nebula with a gas shell of given composition and a compatible central star. The photoionisation model of Aller et al. (1979) finds H+ >99% ionised throughout the nebula and the model of Torres-Peimbert et al. (1990) (their model B) indicates that only 7% of the Lyman continuum is absorbed by the nebula. From the spectra of the central star and NLTE model atmosphere fits, Mendez et al. (1992) determined that 11% of the H Lyman continuum is absorbed by the nebula.

None of these earlier studies had significantly detected low- ionisation species, such as [O II] or [N II], but while the [N II] image in Fig. 5 shows some extended low-ionisation emission, there is no definitive evidence of a classical Strömgren sphere expected for an ionisation bounded nebula. The He I and He II images (Fig. 6, upper right) show that He I is very weak throughout the nebula and He⁺ /H⁺ integrated over the whole nebula is only 4.5% (Sect. 4.2.2) of the total He/H⁺. This raises the question whether NGC 4361 is unusual in also being, at least partially, optically thin in the He II ionising continuum (shortward of 228 Å). The ratio map He I/He II (Fig. 6) shows a marginal radial enhancement, and for the offset field (NGC 4361 W, Table 1), in which emission is detected to radii of ≳40″, He I emission remains very weak. From the integrated emission of all the freckles, the He⁺/H⁺ ratio is 0.11 (see following section), whilst for the extended nebula (excluding the freckles) the He⁺/H⁺ = 0.004 and He⁺⁺/H⁺ = 0.095, so that the total He/H⁺ for the nebula is close to the value for the freckles, i″mplying that there is little margin for extra He⁺/H⁺ (see Table 11 and discussion in the following section).

It was considered whether the optical thinness arises from overall low density or an inhomogeneous structure with large low density opening angles, or a combination of both. The spatially extended (i.e. excluding the freckles) Ne values from [S II] 6716/6731 Å and [Cl III] 5517/5537 Å ratios indicate a mean value (weighted by inverse errors) of 1580 cm⁻³. Torres-Peimbert et al. (1990) quote a total dereddened Hβ flux (corrected for contamination by the He II 4859.3 Å n=4–8 line) of 3.63×10⁻¹¹ erg cm⁻² s⁻¹ in an 81″ diameter aperture; compared to the dereddened [c = 0.10] integrated Hβ flux in the central MUSE field (diameter 68″) of 1.97 × 10⁻¹¹ erg cm⁻² s⁻¹ (also corrected for He II 4859 Å), thus confirming that flux was lost through lowered atmospheric transparency and the non-optimal sky subtraction (Sect. 2), in addition to the smaller aperture. For an assumed spherical nebula of diameter of 80 arcsec⁴, and using the Hummer & Storey (1987) Hβ emissivity for the H I T_PJ of 7500 K (Table 5, column 5), the root mean square density is 190 cm⁻³, implying a filling factor (Torres-Peimbert et al. (1990), Sect. 6) of 0.015, for an electron yield per atom of 1.25 (since 95% of He is in the form of He⁺⁺). The very low filling factor suggests that the line of sight is filled by only a small volume of gas, whose density is the same as the value determined from the CEL diagnostic ratios (viz. [S II] and [Cl III]). From the appearance of the nebula, a possible morphology could be a barrel whose rim has the density given by the diagnostic CEL lines but whose centre has a much lower density and whose ends (to the NE and SW where the ‘ears’ are seen) are open to H Lyman continuum radiation escape. The kinematic mapping of Muthu & Anandarao (2001), interpreted as a bipolar (or even quadrupolar) form, would at least suggest a structure whose NE lobe is tilted towards the line-of-sight, and SW one away from the line-of- sight, as also implied by the large-scale trend in the Hα velocity image (Fig. 10).

The optical colour temperature of the central star indicates a BB temperature ~89 kK, whilst fits to Rauch (2003) model atmospheres indicate values ~80 kK (Appendix A). These values are close to, and certainly not exceeding, the He II Zanstra temperature estimates, indicating no strong evidence for the escape of He II ionising photons (i.e. underestimate of the number of photons between 13.6 and 54.4 eV). Nevertheless, the extreme He II/Hβ line ratio of NGC 4361 among PNe, the very high ionisation level of the bright emission lines and the weakness of the He I emission (except in the freckles), suggest that the nebula could be optically thin in the He⁺ ionising continuum, at least in some directions. Only a detailed photoionisation model with a tuned CS model atmosphere together with deeper spectroscopy of the outer extent of the nebula, could resolve this question.

6.2 Properties of the NGC 4361 freckles

The histograms presented in Fig. 11 summarise some of the essential properties of the freckles:

• the [N II]/Hα ratio (mean value 0.45) far exceeds the ratio for the integrated high-ionisation nebula (0.0013), which led to their detection on the [N II] image (Fig. 5);

• the extinction is generally larger than for the large-scale nebula (mean value: 0.39);

• the electron density is more often lower than the value determined for the nebula overall, also the mean value (N_e = 1160 cm⁻³);

• the T_e ([N II]) is much lower (mean value 13 200 K) than the equivalent value for the total nebula (excluding the freckles) of 27 700 K;

• the T_e ([S III]) is much larger (mean value 27 100 K) than the equivalent value for the total nebula of 17 100 K.

Concerning the sizes of the freckles, they are not spatially resolved. The apparent size of the freckles in the [N II] 6583 Å image (Fig. 5) does not differ from the point spread function, as measured for the central star at the same wavelength based on 2D Gaussian fits to nine well isolated and obviously single freckles (full width at half maximum, 0.69±0.04″, compared to the PSF of the CS, 0.73″, at 6600 Å). Thus the dimension of the freckles is <10¹⁶ cm.

A comparison of the abundances of the light elements (He, N & O) between the freckles and the bulk nebula could strengthen indications that the freckles represent a distinct component. Based on the T_e and N_e values presented in Table 8 for the summed freckle (A+B) spectra (adopting T_e ([N II]) 11 000K and N_e ([S II]) 1500 cm⁻³) and values of T_e 17 000 K and N_e ([S II]) 1500cm⁻³ from Table 4 for the high-ionisation medium (Sect. 6.1), Table 11 presents He⁺, He⁺⁺, N⁺, and O⁺, O⁺⁺, S⁺, S⁺⁺, and total He/H⁺ , N/H⁺ , O/H⁺, S/H⁺ and N/O abundances. The total abundances for N, O, and S, were calculated with the ionisation correction factors (ICFs) of Kingsburgh & Barlow (1994) and Delgado-Inglada et al. (2014), but it should be borne in mind that both scales are developed on lower ionisation PNe and so the ICF’s will not be accurate for the integrated high-ionisation component of NGC 4361 (for example, the ICF for N is 329 based on Kingsburgh & Barlow (1994)). The value of N/O should be more reliable than N/H⁺, and shows that the freckles have a higher value than the bulk nebula gas; the values of He/H⁺ are not significantly different between both components, given that the error on He⁺/H⁺ for the freckles is ∽ 10%. However given that N/O derived for the high-ionisation component is very low compared to typical Galactic Disk PNe (Kingsburgh & Barlow 1994), while the value for the freckles is closer to the current Solar value (Amarsi et al. 2021), suggests problems in determining N/O for such an extreme ionisation medium. The O/H⁺ value is also larger for the summed freckles than for the bulk nebula by about a factor 2 (depending on choice of ICF); this increase hints at the O abundance of the freckles being larger, but again must be treated with caution given that the ICF’s are being applied beyond their recommended range (in He⁺⁺/He). S/H⁺ is also higher for the freckles than the bulk nebula with the ICF from Kingsburgh & Barlow (1994), but lower with the ICF from Delgado-Inglada et al. (2014).

Another aspect of the freckles that distinguishes them from the extended high-ionisation nebula, is their radial velocity. Figure 10 shows the freckle velocities on the large-scale Hα velocity field, whose main feature is a gradient from E to W (∽−12 to +20 km s⁻¹ with respect to systemic velocity, or −15 to +24 km s⁻¹ NE to SW). Most freckles show a positive or negative velocity offset with respect to Hα (Table 6, column 7). Examining the image of the velocity difference ([N II] − Hα) reveals that the freckles segregate between +ve offset to the NE and −ve offset to SW, with only 12/102 within ±5 km s⁻¹ of the zero velocity (= local Hα velocity). The line demarcating the −ve from +ve freckle velocity offsets is at PA∽ 136°, thus approximately perpendicular to the large-scale ellipticity of the PN morphology (PA ∽ 44°). This strongly suggests that the system of freckles forms a thick equatorial structure, perhaps a disc, around the more diffuse high-ionisation core, with the NE portion titled away from the observer (+ve freckle velocities) and the SW towards the observer into the plane of the sky (−ve velocities).

However this picture is complicated by the faint extended [N II] structure which is visible in the 6583 Å image (Fig. 5) composed of a (mirror inverted) comma-shaped cloud to the E & NE and a more elliptical one to the NW. On the ([N II] − Hα) velocity difference image, the eastern cloud is velocity offset −ve (∽−8 km s⁻¹) and the NW cloud offset in +ve velocity (∽+19 km s⁻¹). This trend is in the opposite velocity sense to the freckles, suggesting it is a different structure, related by similarity to the large-scale nebula, shown in the Hα velocity field in Fig. 10. It is plausible that these regions are lower ionisation layers outside the high-ionisation zone, and indeed these features are seen on higher ionisation CEL images, such as [O III] (Fig. 4, lower left) and the He I 5876Å/He II 5412 Å (Fig. 6, upper right). Partial recombination of He⁺⁺ is occurring in these regions implying increased optical depth to He⁺ ionising photons in some parts of the nebula, perhaps in outer extremities of the structure. However the putative disc of freckles does not show up as extended emission and as mentioned has a distinctly different velocity displacement. North and South of the CS, where there is increased spatial density of freckles, extended low-ionisation emission is very weak.

6.3 Nature of the NGC 4361 freckles

The freckles present a radically dichotomous nebular phase to the bulk very high-ionisation medium, and as a result their nature is mystifying; the most extreme difference is seen in the factor ∽340 for [N II] 6583 Å/Hβ ratio between freckles and the integrated high-ionisation emission. They also present a distinct component to the extended low-ionisation emission regions to the east, NW and SW (Fig. 5), which also have similar morphology to the brighter mid-ionisation (30–60 eV) CEL emission (Fig. 4). Their distinction from the other nebula components raises the question whether the freckles could be remoter from the CS than the extended CEL layers. They may represent late ejections from the star or the remains of a distinct component of the asymptotic giant branch (AGB) envelope, such as a disk, perhaps even from an earlier (pre-AGB) phase, such as a circumstellar disk with the freckles as planets or asteroids. Their spatial distribution and trend in velocity from -ve SW of the CS to +ve to the NW, suggests a distribution in a thick disc oriented almost perpendicular to the extension of the highly ionised, optically thin component. This orientation also coincides with ’reflection axis B’ in the quadrupolar interpretation of Muthu & Anandarao (2001). However the velocity tilt of the disc along a NE-SW axis is distinct from the velocity trend mapped by Muthu & Anandarao (2001), suggesting a sextupolar structure in their picture.

It is tempting to interpret the radial orientation of some knots as indication of a collimated ejection origin, but only a minority of the freckles can be fitted into this picture – so either the freckles are a mix of density peaks within a larger (disc) structure or later ejection knots, with the eye led to over-interpret almost linear associations as connected features (particularly since linear trends of knots do not display distinct velocity gradients). Higher velocity resolution spectra and an investigation of any proper motions (a long term goal given that the freckles have only been detected in the epoch 2014 observations) would be required to advance these suggestions. The increased N/H⁺ for the freckles, and more uncertainly He/H⁺, would tend to favour a distinct ejection, perhaps associated with a later evolutionary phase, but the uncertainty in N/O determination for the high-ionisation gas weakens this claim.

The freckles are suggested to be condensations within a larger equatorial disc structure, remnants of dense regions which have been photo-evaporated and are currently ionised. However from the lack of high Ne for the freckles, it is not clear that they are high density condensations, but the increased extinction, the low ionisation, and presence of neutral emission ([O I]) point in the direction of optically thick clumps; then the low [S II] Ne could be a measure only of the ionised flow from a dense globule. Also the trend of higher T_e ([S III]) than T_e ([N II]) could suggest photo-evaporated gas which becomes more highly ionised (and excited) further from the condensation, approaching the higher ionisation optically thin conditions.

From their properties with respect to the bulk nebular emission (low ionisation, elevated extinction, and lack of high Ne , similar abundances), the freckles resemble the low-ionisation structures (LIS) investigated by Akras & Gonçalves (2016) and Mari et al. (2023). However, the freckles show a much higher spatial frequency than, for example, LIS’s in NGC 7009 and NGC 6543 (Mari et al. 2023), are more compact, show enhanced T_e ([N II]) and occur in an older PN⁵.

It is therefore more probable that the freckles are analogous to the cometary knots in the Helix Nebula (NGC 7293) and the extinction knots in NGC 6720 (Wesson et al. 2024), rather than to LIS’s in younger PN. While the extinction to the freckles is generally higher than the surroundings, extinction cores to the freckles are not detected in the [O III] images (or for higher ionisation lines) even for freckles with −ve velocities (presumably in the foreground), as in the case of the cometary knots in the Helix Nebula, NGC 7293 (Meaburn et al. 1992; O’Dell et al. 2000), or similar extinction knots in NGC 6720 (O’Dell et al. 2013). NGC 4361 is however about five times more distant than the NGC 7293. Again higher velocity resolution observations could reveal more details of the ionised structure of the knots and also imaging at higher resolution could test if the knots show structural similarities to those in NGC 7293 and NGC 6720.

The freckles disc could be remoter from the CS than the extended CEL layers hinting at an earlier evolutionary status, perhaps related to the preceding AGB phase. If the low ionisation of the freckles were a result of their being very remote from the CS, in a zone where the nebula is optically thick, it is tempting to suggest they could be similar to planets/asteroids. For freckle 94, 26.9″ from the CS, a projected distance ≳0.14 pc, this would rather be equivalent to an Oort Cloud body (2 ×10³ − 2×10⁵ AU) rather than a planet in the Solar System context. Given the largest systemic radial velocity of the freckles is ~40 km s⁻¹, then a lower limit on the age for a late phase ejection is ≳2700 yr, generally less than the kinematic ages derived for various velocity components by Vázquez et al. (1999); Muthu & Anandarao (2001). However if the measured radial velocities of the freckles correspond to an ionised flow from a dense neutral condensation, and not of the globule itself, then any age estimate is biased.

7 NGC 4361’s galaxy in hiding

When panning through all the wavelength slices of the MUSE cube, a small region showing an emission line at 6703 Å was noticed, which does not match any bright PN nebular line. Other lines were found in the same area and their separations clearly showed a set of typical nebular emission lines (Hα, Hβ, [O III] 4959, 5007 Å, [N II] 6548, 6563 Å and [S II] 6717, 6731 Å) with Hα as the brightest. There is thus a background emission line galaxy at this position (25.6″ offset from NGC 4361 CS at PA 267°) ’shining’ through the emitting gas from the PN. Subsequent to this finding, it was noticed that in the Spitzer IRAC two colour image shown in Fig. 1 there is a hint of a bluer area directly W of the central star at a similar position to the galaxy’s line emission. We investigated the nature of this previously hidden galaxy, which is designated as NGC4361-BgGal1224290-184707⁶.

While the observed emission lines from this galaxy do not coincide with the NGC 4361 nebular lines, the strong continuum (bf, ff, and 2-ν) poses a challenge to extracting its spectrum. The nebular continuum is similarly strong to the galaxy continuum, so must be removed to investigate its stellar properties, such as luminosity, radial profile and stellar population. The spatial and wavelength behaviour of the nebular continuum distribution was fitted by a bicubic in both λ and offset from the central star and scaled by the normalised Hα image of NGC 4361. Only the median continuum flux in steps of 100 Å was fitted over the full cube and regions around the CS and the galaxy itself were excluded by a mask. Figure 12, left image, shows the resulting red-green-blue (RGB) tagged image of the galaxy after this foreground continuum removal. Johnson V and Cousins I photometry was performed on the cleaned spectra of the galaxy using the mpdaf⁷ package (Piqueras et al. 2019) and the radial surface brightness, with the area of the H II regions excluded, was fitted by a Sersic profile (Ciotti & Bertin 1999). Figure 12, right image, shows the expanded view of the ‘blue’ (4850–5850 Å) continuum of the galaxy.

From the Hα image, the H II regions were identified by using hierarchical clustering (dendrogram) models with the Python astrodendro⁸ package. 39 HII regions were thus identified, rejecting any with equivalent circular diameters below the seeing limit (0.7″, or 16 spaxels area). Figure 12, middle image, shows the Hα image of the galaxy. The ionised gas properties of the H II regions were determined by Gaussian fitting the cube over the region of the galaxy; most lines were well separated from the (much brighter) NGC 4361 lines with the exception of Hβ which is within 6 Å of [O III] 4959 Å. Table 12 lists the HII region positions (as offsets from the galaxy centre – see Table 13), fitted radii from astrodendro, Hα luminosities (for a Hubble distance of 87.3 Mpc) and radial velocities, with repect to the systemic velocity of the galaxy of 6416 km s⁻¹ (Table 13). Table 14 then lists the fluxes of the brightest detected lines ([O III] 5007 Å, Hα, [N II] 6583 Å and [S II] 6716 Å) from Gaussian fits to the integrated spectrum of each H II region with errors. The line fluxes have been dereddened by the foreground extinction of NGC 4361. The metallicity, Z, was determined from the O3N2 calibration of Marino et al. (2013) for those (14) H II regions where the lines required for the O3N2 calibration ([O III]5007 Å Hβ⁹, [N II]6583 Å and Hα) were detected.

Subtracting the mean PN extinction (see Sect. 4.1), the extinction to the galaxy was determined (assuming minimal Galactic extinction beyond the PN); for most H II regions the value is compatible with zero but for the brighter central region (ID. 17 in Table 12) E_B−V = 0.20 (c = 0.14) was found. In some H II regions the [S II]6716/6731 Å ratio allowed Ne to be measured, but errors are large and the results are compatible with the low density limit. The radial velocity of the galaxy was then determined as the mean of the velocities of these H II regions (listed in Table 13), and their galactocentric radial velocities indicate a rotation amplitude of ~140 km s⁻¹ with the SW side receding.

Table 13 lists the properties of this previously hidden galaxy. NGC4361-BgGalJ1224290-184707 has a projected size of ~ 14 kpc making it a similar size to NGC 300 (but note that the whole galaxy is not covered in the deep central MUSE field), while it is comparable in luminosity to the Small Magellanic Cloud (M_V −17.1, de Vaucouleurs et al. (1991)) and mass (Besla 2015). The bulge appears to be much brighter than the disk, however, suggesting that it is likely an Sa or Sb galaxy (barred or unbarred), as these have more prominent bulges. In summary, it is a low-mass spiral or a Magellanic irregular at ~87 Mpc serendipitously viewed through the nebular shell of NGC 4361.

Fig. 12 RGB image (R: 7500–8500 Å; G: 5600–6600 Å; B: 4850–5850 Å) of the whole MUSE field of view after a NGC 4361 nebular continuum subtraction (left). The continuum emission of the background galaxy is clearly seen to the west. Hα image of the z = 0.021 background galaxy to NGC 4361 (middle). The contours of NGC 4361 Hβ emission are as in Fig. 2. Broadband ‘blue’ (4850–5850 Å) image of the galaxy after NGC 4361 nebular continuum subtraction and foreground extinction correction, again with NGC 4361 Hβ contours (right).

8 Conclusions

NGC 4361 was observed during MUSE Commissioning in 2014. In total 76 MUSE exposures were obtained (amounting to 4560s) and preliminary examination revealed that the quality of most of the data was good. The reduced data have opened up new perspectives on this enigmatic PN.

• NGC 4361 shows an exceptionally high level of ionisation as demonstrated by the He II/H I line ratios, and had been classified as an optically thin (density bounded) PN. Prior to this study no low-ionisation CEL species had been detected. New evidence that the nebula is optically thin in the H ionising continuum comes from the very low He I emission. The lack of decrease in He⁺⁺/H⁺ with radius raises the question if the medium is, at least partially, optically thin in the He⁺⁺ ionising continuum also. Perhaps deeper imaging in the outer regions could provide indications that He II recombines at the extremities

• This study is the first to find the expected, but very weak, low-ionisation emission, both as extended emission in some regions of the nebula and, most spectacularly as a family of compact (unresolved at 0.73″ resolution) knots, dubbed ‘freckles’. The freckles are most prominently detected in [N II], but also in [O II] and some in [O I]. 102 were detected in this study, but deeper imaging may reveal many more.

• A map of T_e from the [S III] line ratio (Fig. 8) shows the regions to the W and NE, which display mid and low-ionisation extended emission, have differing T_e with a differences up to ~4000 K, actually larger than found by Liu (1998) from a long slit placed 10″ E of the CS. This large- scale temperature difference in a PN is very distinct from the usual temperature differences, such as those measured by different diagnostic ratios, and the temperature fluctuation effects observed in many PNe (such as between CEL and ORL diagnostics, see e.g. Liu et al. (2006)). Furthermore, N_e was determined from [Cl III] and [S II] line ratios, with values around 1500 cm⁻³ (Table 4).

• The freckles were studied in some depth and they are shown to behave much like a typical mid-ionisation medium with T_e ~ 11 000 K (from the diagnostic [N II] ratio) and N_e ~ 1500 cm⁻³ . They generally show larger extinction than the high-ionisation nebula, indicating a distinct dust component, but not significantly higher N_e. Sections 5.2 and Table 7 present the extracted spectra of the freckles. The lower ionisation of these knots is very striking and, for example, the He/H abundance of ~0.1 is wholly due to He⁺, whilst in the high-ionisation gas the majority is He⁺⁺ . There is no compelling evidence that the freckles are enriched in either He or N, with respect to the high-ionisation medium; [O/H] for the freckles is higher than values found for the high-ionisation medium from earlier studies. Comparisons to low-ionisation knots in other PNe, such as the cometary globules in the Helix Nebula (NGC 7293), do not reveal any similar structure (compact extinction core and ionised head and tail), but the ~5× greater distance of NGC 4361 does not allow for any further resolution of their structure based on these data.

• The spatial distribution of the freckles with several groups of almost linear alignments welcomes the suggestion that they are made up of ejecta from the CS, but no convincing alignments back to the CS have been found. The low spectral resolution of MUSE only offers access to limited data on the velocity structure of the freckles; however, there is still a clear trend between those freckles to the west of an axis at PA ~ 140° (blue-shifted with respect to the bulk emission) and those to the east (red-shifted). The picture is advanced that the freckles are aligned in a thick disk, perhaps of larger radius than the large-scale high-ionisation structure (NE–SW orientation) and with perpendicular orientation. The freckles could be part of an older structure, for instance, coming from the AGB phase (especially as they are dusty) or from an earlier epoch. High spectral resolution would help to elucidate their velocity structure and relation to the bulk high-ionisation gas.

• In the western lobe of the nebula, an emission-line galaxy viewed through the nebula emission was found on panning through the MUSE cube. Further investigation showed this to be a typical H II region galaxy and the redshift was 0.0214. Carefully extracting the galaxy spectrum from the bright nebula emission, a low-luminosity (M_V – 17.3 mag.) disk galaxy with bright H II regions was revealed. The gas phase metallicity is [O/H] ≅ 8.4 and it was suggested to be a Magellanic irregular or low-mass spiral galaxy at 87 Mpc.

Data availability

Full Tables 1, 2, 6, 7, 9, 12, and 14 are available at the CDS via anonymous ftp to cdsarc.cds.unistra.fr (130.79.128.5) or via https://cdsarc.cds.unistra.fr/viz-bin/cat/J/A+A/690/A264

Acknowledgements

We thank the MUSE Team for boldly taking the observations of this unusual and less-studied PN during the instrument commissioning time. We are very grateful to the referee for comments, particularly on the quality of the freckles spectra, that led to focussing and improvement of the manuscript. This work has made use of data from the European Space Agency (ESA) mission Gaia (https://www.cosmos.esa.int/gaia), processed by the Gaia Data Processing and Analysis Consortium (DPAC, https://www.cosmos.esa.int/web/gaia/dpac/consortium). Funding for the DPAC has been provided by national institutions, in particular the institutions participating in the Gaia Multilateral Agreement. This research has greatly benefited from the use of PyNeb (Luridiana et al. 2015), which has vastly homogenised calculations of nebular physical conditions and abundances. We are also grateful to the communities who have developed the many Python packages used in this research, such as MPDAF (Piqueras et al. 2019), Astropy (Astropy Collaboration 2013, 2018, 2022), numpy (Walt et al. 2011), scipy (Jones et al. 2001) and matplotlib (Hunter 2007).

Appendix A Spectrum of the central star of NGC 4361

The spectrum of the central star of NGC 4361 was extracted over an area of 42.2 arcsec² (effective radius 3.7″) chosen to include ∼91% of the V-band stellar flux (since the local nebula background varies spatially, the actual flux loss is uncertain). Four rectangular background areas NE, SE, SW, NW of the CS were chosen to avoid the prominent cardinal scattering pattern of the point spread function, and the background subtracted stellar spectrum was formed by subtracting the mean background value from the star image. Figure A.1 shows the resulting spectrum, dereddened by the mean value for the integrated nebula (c=0.10). Both absorption (H I and He II) and emission lines are detected and Table A.1 lists the reliably identified lines by matching with the Atomic Line List of van Hoof (2018). Note that C IV is present in the stellar spectrum and as extended emission with a morphology similar to the He II emission.

A black body was fitted to the dereddened stellar spectrum in Fig. A.1 and a value of 89000 K is indicated. However beyond 8200 Å the black body fit deviates from the dereddened continuum flux: this could arise from flux calibration issues (the flux standard GD 108 (Bohlin et al. 2014; Oke 1990) was used with the observations) and/or aperture losses associated with the broader MUSE PSF extended to longer wavelengths. A match with the hot model atmospheres of Rauch (2003) was also sought. Mendez et al. (1992) lists a model atmosphere fit with T_eff 82000 K and log g 5.5 from non-LTE model atmosphere analysis. The Rauch (2003) T_eff 80000 K H+He models (log g 5.5) fit the continuum well but for a 10% He atmosphere both the Hβ and He II 5412 Å absorption model line profiles are too narrow and shallow (Fig. A.1). A better match in terms of line width and absorption depth to these two lines is found for a 20% He atmosphere, which however considerably exceeds the He/H abundance measured for the nebula and expected for a typical PN central star. For the 90000 K, log g 5.5, 10% He model, the fit to the Hβ profile shape is good but the line too deep, while the He II line fit is to weak and narrow. The fit for the He II line again improves for a 20% He atmosphere but is still slightly too weak and narrow. A value of T_eff of around 85000 K and log g of 5.5 with a 10–15% He contribution appears to be an adequate description of the CS matching this spectrum. However, the much higher T_e ff of 126kK listed by Ziegler et al. (2012) remains puzzling.

Fig. A.1 Dereddened spectrum (c=0.10) of the central star (CS) of NGC 4361 is shown at the top, together with a black body fit (T = 89 000 K, in red) and a Rauch (2003) model atmosphere for Teff 80 000 K, log g = 5.5 and H:He = 0.9:0.1 in green. Bottom panel shows details of the matches of the BB fit and the model atmosphere match for Hβ (left), He II 5412 Å (middle) and Hα (right).

Appendix B Atomic data used in PyNeb diagnostic CEL line ratios and abundances

Table B.1 lists the sources for the collision strengths and transition probabilities used for the PyNeb diagnostic CEL ratios and the abundance calculations.

References

All Tables

All Figures

	Fig. 1 Location of the central MUSE field is shown, superposed on an image of the wider field including the nebula extent, from the Spitzer IRAC combined 3.6 and 4.5 µm image. The scale and orientation are indicated. Image credit: NASA/JPL-Caltech.
In the text

	Fig. 2 Images of NGC 4361 in Hβ, He I 5876 Å and He II 5412 Å with linear colour table lookup (see the colour bar right of each figure). The two coloured wedges at the SE and NW corners of the field shown on the Hβ image represent the regions used for sky background estimation. The contour map on the images is derived from the Hβ image.
In the text

Fig. 3 Selection of representative spectra (log10 flux) of a few given regions. Top: full observed spectrum over a bright region in the core of the nebula (area 1 arcsec2 − 25 MUSE spaxels) centred at (Δα − 5.2″, Δδ + 6.4″) relative to the position of the CS. Bottom: three example spectra over freckles chosen to sample faint, average, and bright freckles (based on their extracted Hα and [N II] fluxes; see Sect. 5.2). Left: spectrum of a faint freckle (9). Centre: Spectrum of a freckle (12) whose Hα and N II fluxes mark it as of average brightness; spectrum of the brightest freckle (94). Note: these spectra of freckles include the (high-ionisation) background nebular spectrum within the area of the freckle.

In the text

	Fig. 4 Images of NGC 4361 in two Ar ions – [Ar III]7136 Å and [Ar V]7006 Å – demonstrating the differing morphology at two ionisation levels for the same element (top). Images in the two strongest lines of medium ionisation species – [O III]5007 Å and [S III]9069 Å (bottom). The contour lines are derived from the Hβ image in Fig. 2.
In the text

	Fig. 5 Strikingly contrasted appearance of NGC 4361 at low ionisation is shown by this image of the brightest optical line [N II] 6583 Å.
In the text

	Fig. 6 Ratio images of NGC 4361 in He lines (top): He II 5412 Å/Hβ (left) and He I 5876 Å/He II 5412 Å (right) demonstrating that a very large fraction of He is fully ionised throughout the nebula. Ratio images displaying the ionisation zones (bottom): [O III] 5007 Å/Hβ (left) for the medium ionisation zone and [Ar V] 7006 Å/Hβ for the highest ionisation (right).
In the text

	Fig. 7 Log extinction at Hβ, c, image of NGC 4361 is shown for the central MUSE field. The area over the bright central star has been substituted by a constant value (see text for more detail). The contour lines are for Hβ flux as in Fig. 2.
In the text

	Fig. 8 Te image from [S III] 6312/9069 Å. The original MUSE spax- els were binned 5×5 (1.0×1.0″) to increase the S/N in the fainter line (6312 Å).
In the text

Fig. 9 Dereddened spectrum of the full MUSE field of NGC 4361, with the area of the central star subtracted, is shown with the fit of the H+, He+ and He++ nebular continua (in blue, yellow and red, respectively). The sum of the three sets of continua is shown in green. The fit to the spectrum is shown in magenta and includes a scattered light contribution modelled as a third-order polynomial continuum to match the sum of the nebular continua to the measured nebular spectrum. See text for further details and Table 5 (Columns 5 and 6) for the fitted jump temperatures (H I Paschen jump and He II n=5 jump).

In the text

	Fig. 10 Velocity field for the central MUSE field NGC 4361 determined from Hα is shown together with the numbering for the freckles, with a circle for each proportional in size to the [N II]6583 Å flux. The Hα and individual freckle offset velocities ([N II] − local Hα), in km s−1, are colour-coded as shown by the colour bar on the right.
In the text

Fig. 11 Histograms of the properties of the freckles: the observed Hα flux (log erg cm−2 s−1) (top left); the [N II]6583 Å/Hα ratio (upper right); log extinction at Hβ, c from Hα/Hβ (centre left); Ne ([S II]) (cm−3 (centre right); Te ([N II]) K (bottom left); Te ([S III]) K (bottom right); The vertical bar in the plots shows the corresponding value for the large- scale nebular emission of NGC 4361 from the MUSE central field, but excluding the area of the freckles. Note: each histogram only plots the available number of values of each parameter (listed in the title to each plot as (#) ); for the extinction, Ne and Te histograms only the values from the A and B quality spectra are shown.

In the text

Fig. 12 RGB image (R: 7500–8500 Å; G: 5600–6600 Å; B: 4850–5850 Å) of the whole MUSE field of view after a NGC 4361 nebular continuum subtraction (left). The continuum emission of the background galaxy is clearly seen to the west. Hα image of the z = 0.021 background galaxy to NGC 4361 (middle). The contours of NGC 4361 Hβ emission are as in Fig. 2. Broadband ‘blue’ (4850–5850 Å) image of the galaxy after NGC 4361 nebular continuum subtraction and foreground extinction correction, again with NGC 4361 Hβ contours (right).

In the text

	Fig. A.1 Dereddened spectrum (c=0.10) of the central star (CS) of NGC 4361 is shown at the top, together with a black body fit (T = 89 000 K, in red) and a Rauch (2003) model atmosphere for Teff 80 000 K, log g = 5.5 and H:He = 0.9:0.1 in green. Bottom panel shows details of the matches of the BB fit and the model atmosphere match for Hβ (left), He II 5412 Å (middle) and Hα (right).
In the text