We present a detailed strong‐lensing, weak‐lensing and X‐ray analysis of Abell 2744 (z= 0.308), one of the most actively merging galaxy clusters known. It appears to have unleashed ‘dark’, ‘ghost’, ‘bullet’ and ‘stripped’ substructures, each ∼1014 M⊙. The phenomenology is complex and will present a challenge for numerical simulations to reproduce. With new, multiband Hubble Space Telescope (HST) imaging, we identify 34 strongly lensed images of 11 galaxies around the massive Southern ‘core’. Combining this with weak‐lensing data from HST, VLT and Subaru, we produce the most detailed mass map of this cluster to date. We also perform an independent analysis of archival Chandra X‐ray imaging. Our analyses support a recent claim that the Southern core and Northwestern substructure are post‐merger and exhibit morphology similar to the Bullet Cluster viewed from an angle. From the separation between X‐ray emitting gas and lensing mass in the Southern core, we derive a new and independent constraint on the self‐interaction cross‐section of dark matter particles σ/m < 3 ± 1 cm2 g−1. In the Northwestern substructure, the gas, dark matter and galaxy components have become separated by much larger distances. Most curiously, the ‘ghost’ clump (primarily gas) leads the ‘dark’ clump (primarily dark matter) by more than 150 kpc. We propose an enhanced ‘ram‐pressure slingshot’ scenario which may have yielded this reversal of components with such a large separation, but needs further confirmation by follow‐up observations and numerical simulations. A secondary merger involves a second ‘bullet’ clump in the North and an extremely ‘stripped’ clump to the West. The latter appears to exhibit the largest separation between dark matter and X‐ray emitting baryons detected to date in our sky.

We present a weak-lensing analysis of the merging Frontier Fields (FF) cluster Abell 2744 using new Subaru/Suprime-Cam imaging. The wide-field lensing mass distribution reveals this cluster is comprised of four distinct substructures. Simultaneously modeling the two-dimensional reduced shear field using a combination of a Navarro-Frenk-White (NFW) model for the main core and truncated NFW models for the subhalos, we determine their masses and locations. The total mass of the system is constrained as M 200c = (2.06 ± 0.42) × 10 15 M . The most massive clump is the southern component with M 200c = (7.7 ± 3.4) × 10 14 M , followed by the western substructure (M 200c = (4.5 ± 2.0) × 10 14 M ) and two smaller substructures to the northeast (M 200c = (2.8 ± 1.6) × 10 14 M ) and northwest (M 200c = (1.9 ± 1.2) × 10 14 M ). The presence of the four substructures supports the picture of multiple mergers. Using a composite of hydrodynamical binary simulations we explain this complicated system without the need for a "slingshot" effect to produce the northwest X-ray interloper, as previously proposed. The locations of the substructures appear to be offset from both the gas (87 +34 −28 arcsec, 90% CL) and the galaxies (72 +34 −53 arcsec, 90% CL) in the case of the northwestern and western subhalos. To confirm or refute these findings, high resolution space-based observations extending beyond the current FF limited coverage to the west and northwestern area are essential.