Numerous bow shocks in the outer Helix Nebula

The Helix Nebula (NGC 7293) is one of the closest and brightest planetary nebulae, and therefore a benchmark for resolving how the late-stage ejecta of stars couple to their surroundings. Using the Gaia EDR3/DR3 astrometric solution for the central star (WD 2226-210), we adopt a distance of d = 198.6 (+1.6/−1.8) pc (ref. ¹⁰). Imaging of the Helix has shown it to be highly complex. Its bright main nebula comprises an inner disc and a surrounding outer torus, embedded within a larger structure whose upstream side is truncated, consistent with interaction between the expanding asymptotic giant branch (AGB) ejecta and the ambient interstellar medium (ISM)^11,12,13. The ionized nebula is threaded by thousands of dense cometary knots and associated molecular material, indicating that much of the ejected material remains in a clumpy, only partially processed phase^11,14. Deep imaging and spectroscopy have also revealed a bow-shock feature in the faint outer halo, in the direction of the nebula’s motion through the local ISM^12,15. Together, these properties make the Helix uniquely suited to place direct constraints on how fragmented stellar ejecta are dispersed and mixed into the ISM: an essential step in the recycling of mass, dust and newly synthesized elements in galaxies^4,9.

The Helix Nebula was observed in the light of Hα, [N ii] and [O iii] with the partially built Modular Optical Telephoto Hyperspectral Robotic Array (MOTHRA). MOTHRA is an array of high-end telephoto lenses equipped with tiltable ultra-narrow interference filters, located at the El Sauce Observatory in Chile. Its design evolved from the Dragonfly Spectral Line Mapper at New Mexico Skies Observatory^16,17. When completed, MOTHRA will have 1,140 lenses distributed over 30 mounts, and be optically equivalent to a 4.8 m f/0.08 refractor. The data described here are equivalent to roughly 20 minutes of on-source exposure time with the completed array.

The MOTHRA Hα image of the Helix is shown in Fig. 1. It shows many features that have not been seen in ionized gas before, such as extensions of the plumes in the northwest and southeast^11,13 and turbulent and complex low surface brightness Hα emission in the southwest¹³. The most striking feature in the Hα image is a forest of arcs and partial arcs on the eastern side of the nebula. We identify at least 22 arcs on the eastern side, labelled 1–22 in Fig. 3 in order of increasing distance from the central white dwarf. Although most of the features appear to be new discoveries, several can be seen in previous GALEX and Hα images. Besides the large and complex feature 14 this includes arcs 3, 10 and 13, among others^11,13,15. The arcs are undetected in [O iii] and faint in [N ii]; from the brightest region of arc 14 we measure [N ii]/Hα = 0.06 ± 0.01 and [O iii]/Hα < 0.015 (2σ). We also find faint arc-like features on the western side, at a similar distance from the white dwarf as the much brighter ones in the east.

Following earlier studies we interpret the eastern features as bow shocks in which expanding stellar ejecta encounter the ISM at supersonic velocities^13,15. The velocity of the Helix with respect to the ISM is roughly 36 km s⁻¹ towards 95° east of north in the plane of the sky and roughly 27 km s⁻¹ along the line of sight^15,18, for a combined ISM velocity of v_ISM ≈ 45 km s⁻¹ with respect to the systemic velocity of the nebula. On the eastern side the shock velocity is the sum of v_ISM and the expansion velocity of the ejecta, v_{shock, e} ≈ |v_ISM + v_exp|. On the western side the ejecta encounter a turbulent postshock wake that has passed through, and mixed with, the planetary nebula, v_{shock, w} ≈ |v_wake − v_exp|. The wake velocity is expected to be small with respect to the systemic velocity of the nebula. In the case of the well-studied AGB star Mira, the processed gas being shed from the bow shock into the immediate downstream tail lags the star by only roughly 10% (ref. ¹⁹); applying the same scaling to the Helix gives v_wake = 0−10 km s⁻¹ for the flow on the western side.

Using the MAPPINGS V code²⁰ we derive shock velocities on the eastern side of v_{shock, e} = 80−90 km s⁻¹ from the [O ii]/Hα and [N ii]/Hα line ratios (Methods). Subtracting the 45 km s⁻¹ ISM velocity gives an expansion velocity of the ejecta of v_exp = 35−45 km s⁻¹, consistent with previously measured Hα kinematics in the region of the brightest bow¹⁵. The implied velocity field around the Helix is shown in Fig. 2. Near the east–west axis the shock velocities are roughly 80 km s⁻¹ in the east and roughly 35 km s⁻¹ in the west. For these velocities, shock models predict Hα luminosities that are 1 to 2 orders of magnitude fainter in the west than in the east, consistent with the appearance of the bows on the two sides of the Helix (Methods).

The expansion velocity of v_exp = 35–45 km s⁻¹ indicates a dynamical age of the clumps of 20,000–30,000 years at r ≈ 1 pc, predating the formation of the planetary nebula roughly 12,000 years ago²¹. The material therefore most likely belongs to an older circumstellar envelope that was ejected during the late-AGB phase, and has now fragmented into many individual clumps. This interpretation is strengthened by the fact that the bows lie at approximately the same radius as the roughly 40′ outer WISE 12-μm halo, which has been associated with dust from an AGB wind¹³. Fast winds and outflows, with velocities that can exceed the canonical roughly 5−20 km s⁻¹ range of the main AGB phase²², are commonly observed during the late-AGB and early post-AGB phases^23,24. These flows are often bipolar rather than isotropic²⁵, which may explain why there appears to be a preferred axis connecting the strong bows in the east to northeast of the Helix to the weak bows in the west to southwest.

We fit the bow morphologies with a range of functional forms: parabolas, hyperbolas, ellipses and Wilkinoids²⁶. We adopt the parabolic fits as our fiducial model, as they provide a reasonable description of the data and retain the same functional form under projection²⁷. As many of the structures do not show a complete bow, and some are broader and less sharply bounded than ideal thin-shell bow shocks, we do not attach direct dynamical meaning to the adopted fit family itself. Instead, we use the fits to extract geometric quantities (in particular the characteristic curvature scale), and use the variation between fit families as an estimate of the systematic uncertainty. The fitting procedure is detailed in the Methods, with the results shown in Fig. 3. Most of the 22 bows are reasonably well fit by a parabola, although the wings are often closer to hyperbolic²⁷. The brightest bow (14) is more sharply peaked than the model curve; inspection of the region near the apex shows that it is broken up in a complex network of smaller shocks.

The foci of the parabolic fits are indicated with red dots in Fig. 3. These are the approximate locations of the shell fragments that produce the shocks, although the exact location depends on the three-dimensional orientation and shape of the bows²⁷. There is generally nothing visible in Hα, [O iii] or [N ii] at or near these locations. The lack of detected emission counterparts at most of the inferred obstacle locations is consistent with the fragments being largely neutral. Our observations thus represent a new observational window on the mixing of AGB and planetary nebula ejecta into the ISM, in which the fragments are identified not by their intrinsic emission but by the shocks that they drive.

The nature of the bows changes systematically with distance from the white dwarf: close to the star they are large, thin and well-defined, whereas in the outskirts they are smaller and fuzzier. To quantify this trend without assuming a particular steady-state bow-shock solution, we characterize each structure by the radius of curvature at its apex R_c, as measured from the best-fitting parabola. The relation between R_c and the distance from the white dwarf, r, is shown in Fig. 4. The characteristic size of the bows decreases by two orders of magnitude over the radial range 0.4 pc ≲ r ≲ 1.4 pc. A log-linear fit gives log R_c = 0.34−1.59r, corresponding to an e-folding length of 0.27 pc.

Because R_c is a purely geometric quantity, it does not by itself specify the detailed momentum balance within the flow. It does, however, show that the spatial scale of the coherent bow-forming obstacle decreases strongly with outward distance. This geometric trend is accompanied by a systematic morphological transition: the inner bows are thin and sharply bounded, whereas the outer structures are broader, more irregular, and increasingly clumpy. Taken together, these changes suggest progressive stripping and fragmentation of the dense AGB-shell remnants as they interact with the ambient medium. As material is ablated from the fragments and mixed into the surrounding flow, the surviving dense heads become smaller and more porous, and a larger fraction of the Hα emission probably arises in fragment-associated, mass-loaded mixed gas rather than in a geometrically thin, well-defined forward shock^8,28,29.

In this interpretation, the bows are both signposts of the fragments and agents of their destruction. The shocks are powered by the relative kinetic energy of the fragments and the ambient flow, and their presence implies ongoing momentum transfer, ablation and mixing^8,28. Interpreting the radial locations of the bows, r_i, as a time sequence for outward evolution, t_i ≈ r_i/v_exp, the slope of the R_c–r relation implies a characteristic timescale for the loss of coherent bow structure. For the expansion velocity derived from the shock velocities in the east, v_exp ≈ 40 km s⁻¹, the curvature scale declines with an e-folding time τ_Rc ≈ 7 × 10³ years. We therefore infer that the bow-forming AGB-shell fragments are disrupted on a timescale of order 10⁴ years. This estimate should be interpreted as the survival time of the coherent dense fragment–bow system, rather than as a direct measurement of a specific momentum or mass-loss rate.

Classical AGB–ISM bow shocks such as Mira trace a single, wind-driven stand-off interaction centred on the mass-losing star⁶, and far-infrared surveys show that such global wind–ISM interaction structures are common around evolved stars⁷. However, these long-lived⁶ (roughly 10⁵ years) large-scale bows primarily map where the wind meets the ISM; they do not directly constrain how fragmented ejecta are ultimately assimilated. In the Helix outer halo we instead resolve numerous compact bow shocks with no luminous source at their foci, indicating that the AGB–ISM interaction has fragmented into dense, line-dark obstacles that are progressively ablated and entrained. The roughly 10⁴ years characteristic disruption time for the fragments can therefore be interpreted as the relevant timescale for recycling of late-AGB ejecta into the ISM.

More broadly, stellar mass loss is a major channel by which galaxies recycle gas, metals and dust back into the ISM³⁰, yet the efficiency and duration of the final, fragment-driven AGB–ISM assimilation step remain poorly constrained observationally^6,31. Galaxy formation simulations therefore rely on subgrid turbulent mixing and diffusion prescriptions to represent unresolved transport^32,33,34. Our empirically inferred roughly 10⁴ years disruption time implies that once AGB ejecta are fragmented and exposed to the diffuse medium they lose their coherent identity rapidly, providing a benchmark for models of recycling and feedback.

Our conclusions can be tested and extended in several ways. The Helix is a fairly typical planetary nebula, and we should see similar fragment-driven bow shocks in the outskirts of other planetary nebulae. Because of the steep relation between Hα luminosity and shock velocity such features will be most readily detected when the bulk motion of the nebula with respect to the ISM exceeds roughly 40 km s⁻¹. When more examples are found, it will be interesting to see if the mixing timescale depends on the shock velocities. Such observations will be within easy reach of the completed MOTHRA. Furthermore, by analogy with the molecular cometary knots in the Helix^14,35,36,37, the Hα-dark clumps whose presence is inferred from the shocks could be detectable in CO rotational lines and, particularly, in the H₂ 1–0 S(1) 2.12-μm line.

Observations

The Helix Nebula was observed as a commissioning target during the early construction phase of MOTHRA. When completed, the array will comprise 1,140 Canon 400 mm, f/2.8, II/III telephoto lenses distributed over 28 narrow-band mounts and 2 broadband mounts. Each narrow-band mount has 38 lenses: 18 are equipped with an Hα filter, 8 with [O iii] λ5007, 8 with [N ii] λ6583 and 4 are continuum filters. The widths of the ultra-narrow interference filters are 0.71 nm for [O iii] and 0.93 nm for Hα and [N ii], corresponding to a velocity bandwidth of roughly 430 km s⁻¹. The continuum filters, two for Hα and [N ii] and two for [O iii], are roughly 90 nm wide with a roughly 33-nm notch in the middle at the location of the emission lines. These filters therefore measure the continuum at the wavelengths of the lines, without being contaminated by them.

The observations reported here were obtained on 16 November and 19–24 November 2025, when the first five mounts of MOTHRA were operational. As the number of active lenses varied during the observations, we express the exposure time as single-lens equivalent hours, that is, the exposure time multiplied by the number of lenses that were in use during the observation. In Hα the on-target exposure time was 172.5 single-lens equivalent hours, corresponding to 20 min with the 504 Hα lenses that the completed MOTHRA will have.

Data reduction

The data reduction includes several steps that are unique to the design of MOTHRA and its prototype, the Dragonfly Spectral Line Mapper^16,17. The roughly 400 km s⁻¹ wide interference filters are placed in front of the lenses, and specific wavelengths are chosen by tilting the filters. Owing to the large (roughly 3°) field of view, the wavelength of the bandpass is not constant over the image but varies as a function of the angular displacement of the target relative to the projected tilt axis of the filter, as well as the angular distance to the optical axis. As a result, the sky background is complex: sky emission lines appear as broadbands in the image, particularly at the large tilts that are used for Galactic objects¹⁶. For the Helix observations, we obtained offset sky exposures in a circular pattern around the science field to model the background. Bracketing sky exposures were combined and subtracted from each individual science frame, before combining the science data.

A colour image of the Helix is shown in Extended Data Fig. 1, created from the [O iii], Hα and [N ii] data. It is well established from narrow-band imaging and line-ratio mapping that the morphology of planetary nebulae varies strongly between different emission lines, with low-ionization structures often enhanced in [N ii] relative to [O iii]^38,39,40. The Helix shows the same behaviour in its central regions and faint outer halo^11,21,41. In the MOTHRA images, the forest of bow shocks is relatively bright in Hα and is faint in [O iii] and [N ii].

The final step in the data reduction is the subtraction of continuum emission from the narrow-band data; this was done by matching the point spread functions of the two images and scaling the continuum image to match the normalization of the narrow-band image. Remaining residuals were filled in with the maskfill code⁴². The depth of the images was determined with the sbcontrast method, which empirically determines the contrast sensitivity on a desired spatial scale⁴³. The reduced and sky-subtracted narrow-band images reached a 1σ depth of roughly 2 × 10⁻¹⁹ erg s⁻¹ cm⁻² in Hα on 1′ scales.

Shock velocity

To interpret the observed emission-line ratios we computed a grid of radiative shock models using the MAPPINGS V code²⁰. The models were run with Solar abundances and included a self-consistent treatment of the photoionizing precursor, in which the upstream ionization and temperature structure were iteratively determined from the radiation field produced in the postshock cooling zone. We adopted a preshock hydrogen density of n_H = 5 cm⁻³, consistent with the fiducial density used in recent MAPPINGS V benchmark calculations⁴⁴, and appropriate for diffuse circumstellar and interstellar environments. The results were insensitive to the choice of n_H, as the lines formed when much higher densities were reached. The magnetic field strength was parameterized through α ≡ B/√n_H (with B in μG and n_H in cm⁻³), which controls the degree to which magnetic pressure limits compression in the postshock cooling zone²⁰.

We explored two representative regimes: a weak-field case with α = 1, which effectively captured the hydrodynamic limit (the results were virtually identical for all values α ≲ 1), and a magnetically supported case with α = 2, for which magnetic pressure significantly reduced the maximum compression. Shock velocities were sampled finely in the range v_shock = 20−115 km s⁻¹ to resolve the transition in ionization structure associated with the onset of [N ii] and [O iii] emission.

The results are shown in Extended Data Fig. 2. The line ratios are a strong function of shock velocity in this regime: the postshock temperature increased as T ∝ v_shock (ref. ²), leading to a rapidly increasing flux of ionizing photons from the cooling zone and its associated precursor²⁰. For [O iii], the onset was abrupt once the radiation field became sufficiently hard to sustain an extended O⁺⁺ zone, whereas lower-ionization species such as [N ii] responded more gradually. The observed upper limit on [O iii]/Hα and measured [N ii]/Hα (the grey bands in Extended Data Fig. 2) indicated a shock velocity of v_shock = 80−90 km s⁻¹ depending on the magnetic field strength, and we use v_shock = 85 ± 5 km s⁻¹ in the main text.

Hα luminosity as a function of shock velocity

There is a clear asymmetry in the brightness of shocks in the outer Helix Nebula, with all strong shocks located on the eastern side. As can be seen in Fig. 1 and Extended Data Fig. 1, there seems to be a corresponding region on the other side of the nebula, showing rounded arcs and bubbles that are only detected in Hα. In Extended Data Fig. 3 we show a side by side comparison of the two regions. Here we ask whether the difference in brightness between the eastern and western shocks can be explained by the difference of the shock velocities.

The right panel of Extended Data Fig. 3 shows the Hα luminosity as a function of shock velocity, from the MAPPINGS V code²⁰. At low velocities, the Hα luminosity rises steeply with shock velocity as the shock progressively ionizes neutral hydrogen. At higher velocities, once the radiative precursor pre-ionizes the upstream gas, the hydrogen ionization fraction approaches unity and the Hα emissivity no longer tracks the mechanical energy flux. Instead, additional energy is channelled into heating and metal-line cooling, resulting in a flattening of the Hα–velocity relation.

Grey bands show the inferred shock velocities, 85 ± 5 km s⁻¹ on the east side and 35 ± 10 km s⁻¹ on the west side (main text). Because of the steep rise of the Hα–v_shock relation in this regime, the Hα luminosity is expected to be 1 to 2 orders of magnitude higher on the east side than on the west side. The difference in the observed surface brightness is a factor of roughly ten, broadly consistent with the shock models.

We note that the Hα features on the west side are probably not related to the high velocity jet in that region, even though they appear to be in front of it in projection (Extended Data Fig. 3). The jet has a velocity of roughly 300 km s⁻¹, measured directly from Hα and [N ii] spectroscopy¹⁵, and if the shocks were driven by the jet they would show strong [O iii] emission.

Fitting of bow shocks

The bows are fit in the following way. A bow is initially characterized by five manually selected (x,y) positions: two marking the end points of the visible arc, one marking the approximate apex and two points in between the end points and the apex. Next, a second-order polynomial is fit to these five points. This fit provides a reasonable approximation of the area of the image where the bow is located. A band of typical width ±20 pixels is defined around the polynomial fit. Within this band the ridge of the bow is identified by determining the local maximum in bins along the polynomial. To enhance contrast, an unsharp-masked version of the Hα image is used to measure the ridge.

With the ridge line measurements in place, a fit to an analytic function is performed. Following earlier work²⁷, we fit four families of functions: parabolas, hyperbolas, ellipses and Wilkinoids²⁶. For each functional form we determine the associated radius of curvature R_c and the related distance between the apex and the object R_o. For a parabola R_o = R_c/2; for a hyperbola and an ellipse R_o = R_c/(1 + e), with e the ellipticity and for a Wilkin profile R_o = 3R_c/5. The five manually selected points are not used in the fit. After the initial fit the band is redefined, now on the basis of the specific analytic function rather than the polynomial, and the ridge line measurements and fit are repeated. The process is illustrated for eight of the bows in Extended Data Fig. 4. The fits are stable for the 22 bows that we identify in the image.

As discussed in the main text, we use the parabola as the fiducial profile, and take the root mean-square ranges in R_c and R_o from the four fits as uncertainties.

The reduced, continuum-subtracted Hα image is available at Zenodo (https://doi.org/10.5281/zenodo.19864496)⁴⁵.

We have made use of standard data analysis tools in the Python environment. Image projections were performed with dfproject⁴⁶, available at https://github.com/DragonflyTelescope/dfreproject.

1. Kwok, S., Purton, C. R. & FitzGerald, P. M. On the origin of planetary nebulae. Astrophys. J. Lett. 219, L125–L127 (1978).

   Article  Google Scholar 

2. Iben, I. & Renzini, A. Asymptotic giant branch evolution and beyond. Annu. Rev. Astron. Astrophys. 21, 271–342 (1983).

   Article  CAS  Google Scholar 

3. Borkowski, K. J., Sarazin, C. L. & Soker, N. Interaction of planetary nebulae with the interstellar medium. Astrophys. J. 360, 173–183 (1990).

   Article  Google Scholar 

4. Wareing, C. J., Zijlstra, A. A. & O’Brien, T. J. Vortices in the wakes of asymptotic giant branch stars. Astrophys. J. Lett. 660, L129–L132 (2007).

   Article  Google Scholar 

5. Wareing, C. J. et al. Detached shells as tracers of asymptotic giant branch–interstellar medium bow shocks. Mon. Not. R. Astron. Soc. Lett. 372, L63–L67 (2006).

   Article  Google Scholar 

6. Martin, D. C. et al. A turbulent wake as a tracer of 30,000 years of Mira’s mass loss history. Nature 448, 780–783 (2007).

   Article  CAS  PubMed  Google Scholar 

7. Cox, N. L. J. et al. A far-infrared survey of bow shocks and detached shells around AGB stars and red supergiants. Astron. Astrophys. 537, A35 (2012).

   Article  Google Scholar 

8. Klein, R. I., McKee, C. F. & Colella, P. On the hydrodynamic interaction of shock waves with interstellar clouds. 1: Nonradiative shocks in small clouds. Astrophys. J. 420, 213–236 (1994).

   Article  Google Scholar 

9. Karakas, A. I. & Lattanzio, J. C. The Dawes Review 2: nucleosynthesis and stellar yields of low- and intermediate-mass single stars. Publ. Astron. Soc. Aust. 31, e030 (2014).

   Article  Google Scholar 

10. Bailer-Jones, C. A. L., Rybizki, J., Fouesneau, M., Demleitner, M. & Andrae, R. Estimating distances from parallaxes. V. Geometric and photogeometric distances to 1.47 billion stars in Gaia early data release 3. Astron. J 161, 147 (2021).

    Article  Google Scholar 

11. O’Dell, C. R., McCullough, P. R. & Meixner, M. Unraveling the Helix Nebula: its structure and knots. Astron. J. 128, 2339–2356 (2004).

12. Meaburn, J. et al. The creation of the helix planetary nebula (NGC 7293) by multiple events. Mon. Not. R. Astron. Soc. 360, 963–976 (2005).

    Article  CAS  Google Scholar 

13. Zhang, Y., Hsia, C.-H. & Kwok, S. Discovery of a Halo around the Helix Nebula NGC 7293 in the WISE all-sky survey. Astrophys. J. 755, 53 (2012).

    Article  Google Scholar 

14. Hora, J. L., Latter, W. B., Smith, H. A. & Marengo, M. Infrared observations of the Helix Planetary Nebula. Astrophys. J. 652, 426–441 (2006).

    Article  CAS  Google Scholar 

15. Meaburn, J., Boumis, P. & Akras, S. The bow-shock and high-speed jet in the faint, 40 arcmin diameter, outer halo of the evolved Helix Planetary Nebula (NGC 7293). Mon. Not. R. Astron. Soc. 435, 3462–3468 (2013).

16. Lokhorst, D. M. et al. Realizing the potential of the Dragonfly Spectral Line Mapper: calibration methods and on-sky performance. In Proc. Ground-based and Airborne Telescopes X, vol. 13094 of Society of Photo-Optical Instrumentation Engineers (SPIE) Conference Series (eds Marshall, H. K. et al.) 130942N (SPIE, 2024).

17. Chen, S. et al. First light with the 120-lens Dragonfly Spectral Line Mapper. Publ. Astron. Soc. Pac. 137, 084103 (2025).

18. Gaia Collaboration. Gaia early data release 3: summary of the contents and survey properties. Astron. Astrophys. 649, A1 (2021).

    Article  Google Scholar 

19. Matthews, L. D., Libert, Y., Gérard, E., Le Bertre, T. & Reid, M. J. The discovery of an H I counterpart to Mira’s far-ultraviolet-emitting tail. Astrophys. J. 684, 603–615 (2008).

    Article  CAS  Google Scholar 

20. Sutherland, R., Dopita, M., Binette, L. & Groves, B. MAPPINGS V: astrophysical plasma modeling code (Astrophysics Source Code Library, 2018).

21. Meaburn, J., López, J. A. & Richer, M. G. Optical line profiles of the Helix Planetary Nebula (NGC 7293) to large radii. Mon. Not. R. Astron. Soc. 384, 497–503 (2008).

    Article  CAS  Google Scholar 

22. Höfner, S. & Olofsson, H. Mass loss of stars on the asymptotic giant branch. mechanisms, models and measurements. Astron. Astrophys. Rev. 26, 1 (2018).

    Article  Google Scholar 

23. Balick, B. & Frank, A. Shapes and shaping of planetary nebulae. Annu. Rev. Astron. Astrophys. 40, 439–486 (2002).

24. Sahai, R. in Post-AGB Objects as a Phase of Stellar Evolution Vol. 265 of Astrophysics and Space Science Library (eds Szczerba, R. & Górny, S. K.) 53–63 (Springer, 2001).

25. Bujarrabal, V., Castro-Carrizo, A., Alcolea, J. & Sánchez Contreras, C. Mass, linear momentum and kinetic energy of bipolar flows in protoplanetary nebulae. Astron. Astrophys. 377, 868–897 (2001).

    Article  CAS  Google Scholar 

26. Wilkin, F. P. Exact analytic solutions for stellar wind bow shocks. Astrophys. J. Lett. 459, L31–L34 (1996).

    Article  Google Scholar 

27. Tarango-Yong, J. A. & Henney, W. J. True versus apparent shapes of bow shocks. Mon. Not. R. Astron. Soc. 477, 2431–2454 (2018).

28. Hartquist, T. W., Dyson, J. E., Pettini, M. & Smith, L. J. Mass-loaded astronomical flows – I. General principles and their application to RCW 58. Mon. Not. R. Astron. Soc. 221, 715–726 (1986).

    Article  Google Scholar 

29. Pittard, J. M., Hartquist, T. W. & Dyson, J. E. Self-similar evolution of wind-blown bubbles with mass loading by hydrodynamic ablation. Astron. Astrophys. 373, 1043–1055 (2001).

    Article  Google Scholar 

30. Leitner, S. N. & Kravtsov, A. V. Fuel efficient galaxies: sustaining star formation with stellar mass loss. Astrophys. J. 734, 48 (2011).

    Article  Google Scholar 

31. Matthews, L. D., Le Bertre, T., Gérard, E. & Johnson, M. C. An H I imaging survey of asymptotic giant branch stars. Astron. J 145, 97 (2013).

    Article  Google Scholar 

32. Wadsley, J. W., Veeravalli, G. & Couchman, H. M. P. On the treatment of entropy mixing in numerical cosmology. Mon. Not. R. Astron. Soc. 387, 427–438 (2008).

    Article  Google Scholar 

33. Shen, S., Wadsley, J. & Stinson, G. The enrichment of the intergalactic medium with adiabatic feedback – I. Metal cooling and metal diffusion. Mon. Not. R. Astron. Soc. 407, 1581–1596 (2010).

    Article  CAS  Google Scholar 

34. Rennehan, D., Babul, A., Hopkins, P. F., Davé, R. & Moa, B. Dynamic localized turbulent diffusion and its impact on the galactic ecosystem. Mon. Not. R. Astron. Soc. 483, 3810–3831 (2019).

    Article  CAS  Google Scholar 

35. Huggins, P. J., Bachiller, R., Cox, P. & Forveille, T. Co in the cometary globules of the Helix Nebula. Astrophys. J. Lett. 401, L43–L46 (1992).

    Article  CAS  Google Scholar 

36. Huggins, P. J. et al. High-resolution co and h₂ molecular line imaging of a cometary globule in the Helix Nebula. Astrophys. J. Lett. 573, L55–L58 (2002).

    Article  CAS  Google Scholar 

37. Matsuura, M. et al. A ‘firework’ of H₂ knots in the planetary nebula NGC 7293 (the Helix Nebula). Astrophys. J. 700, 1067–1077 (2009).

    Article  CAS  Google Scholar 

38. Balick, B. Evolution of planetary nebulae. I. Structures, ionizations, and morphological sequences. Astron. J. 94, 671–678 (1987).

    Article  CAS  Google Scholar 

39. Corradi, R. L. M., Manso, R., Mampaso, A. & Schwarz, H. E. Unveiling low-ionization microstructures in planetary nebulae. Astron. Astrophys. 313, 913–923 (1996).

    Google Scholar 

40. Gonçalves, D. R., Corradi, R. L. M. & Mampaso, A. Low-ionization structures in planetary nebulae: confronting models with observations. Astrophys. J. 547, 302–310 (2001).

    Article  Google Scholar 

41. Van de Steene, G. C. et al. Herschel imaging of the dust in the Helix Nebula (NGC 7293). Astron. Astrophys. 574, A134 (2015).

    Article  Google Scholar 

42. van Dokkum, P. & Pasha, I. A Robust and simple method for filling in masked data in astronomical images. Publ. Astron. Soc. Pac. 136, 034503 (2024).

43. Keim, M. A. et al. Tidal distortions in NGC1052-DF2 and NGC1052-DF4: independent evidence for a lack of dark matter. Astrophys. J. 935, 160 (2022).

44. Jin, Y. & Raymond, J. Dialog concerning the two shock codes. Astrophys. J. 989, 203 (2025).

45. van Dokkum, P. MOTHRA continuum-subtracted Halpha image of the Helix Nebula. Zenodo https://doi.org/10.5281/zenodo.19864496 (2026).

46. Rhea, C. et al. dfreproject: a Python package for astronomical reprojection. J. Open Source Softw. 10, 8525 (2025).

    Article  Google Scholar 

Download references

Authors and Affiliations

1. Dragonfly Focused Research Organization, Santa Fe, NM, USA

   Pieter van Dokkum, Roberto Abraham, William P. Bowman, Seery Chen, Steven R. Janssens, Deborah M. Lokhorst, Imad Pasha & Carter Rhea

2. Department of Astronomy, Yale University, New Haven, CT, USA

   Pieter van Dokkum

3. Department of Astronomy and Astrophysics, University of Toronto, Toronto, Ontario, Canada

   Roberto Abraham

4. NRC Herzberg Astronomy and Astrophysics Research Centre, Victoria, British Columbia, Canada

   Deborah M. Lokhorst

Contributions

P.v.D., R.A., W.P.B., S.C., S.R.J., D.M.L., I.P. and C.R. contributed to the construction of the array, the execution of the observations and the data reduction. P.v.D. led the analysis and wrote the paper. R.A. aided in the interpretation.

Corresponding author

Correspondence to Pieter van Dokkum.

Competing interests

The authors declare no competing interests.

Peer review information

Nature thanks William Henney and the other, anonymous, reviewer(s) for their contribution to the peer review of this work. Peer reviewer reports are available.

Open Access This article is licensed under a Creative Commons Attribution-NonCommercial-NoDerivatives 4.0 International License, which permits any non-commercial use, sharing, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if you modified the licensed material. You do not have permission under this licence to share adapted material derived from this article or parts of it. The images or other third party material in this article are included in the article’s Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article’s Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://creativecommons.org/licenses/by-nc-nd/4.0/.

Reprints and permissions