Entry 108 closed with “standoff-first, because nobody can touch a tumbling unprepared object.” The natural next question: touch what, exactly? What is the target made of, mechanically? The sweep’s answer is the most quotable of the series: the canonical rubble pile has the cohesion of a whisper, the canonical small body spins like a top, and every spacecraft that ever made contact discovered the surface was a fluid wearing a rock costume.

The material verdict, in pascals

Cohesion is the number that decides everything about touching, and the measured values are absurdly small. Bennu’s interior holds together at ~1 Pa (a 3.5-hour spin would shred it). Its near-subsurface — the part OSIRIS-REx actually touched — measured 0.2–20 Pa, with the excavation geometry requiring under 1 Pa. Ryugu’s surface strength: <1.3 Pa (from the SCI crater). Dimorphos: probably a few Pa, and DART’s modeling says anything over 4 Pa would have changed the cratering regime entirely — the observed outcome rules it out. Theory (Sánchez & Scheeres) explains the whole range with van der Waals “cement” between grains: 25–100 Pa, scaling inversely with grain size. For calibration: 1 Pa is the pressure a sheet of paper exerts on a table. The keeper’s targets, in the size range where gravity stops compacting anything (packing fraction 0.2–0.45 at Bennu’s surface — looser than the grains naturally settle), are held together by the same force that makes dust stick to a camera lens.

The touch record is a violence ladder

Every contact ever made, in ascending order of energy, and every one exceeded expectations:

  • 10 cm/s (OSIRIS-REx TAG): the arm sank half a meter, peak resistance 10–15 N — Lauretta’s phrase is “squeezing the plunger on a French press,” and the honest counterfactual is that without the backaway burn the spacecraft would have sunk into Bennu. Excavation to 68 ± 10 cm, an ~8 m crater, and so much material mobilized that rocks wedged the sample head open and the mission nearly lost the cargo to space.
  • 1.5 m/s (NEAR at Eros, 2001): survived, only because Eros is the coherent-body case.
  • 2 km/s, 2 kg copper (Hayabusa2 SCI): a >10 m crater — gravity-dominated, meaning the surface contributed essentially nothing to stopping it — and 200+ boulders displaced up to 40 m away, by seismic shaking alone. Hayabusa2’s own RCS plume moved surface boulders during low passes. Even exhaust gas is a contact event.
  • 6.6 km/s (DART): no conventional crater at all — Dimorphos was globally deformed, axis ratio driven measurably more elongated, plausibly knocked into tumbling rotation, and 37 boulders (up to 7 m) drifted off the surface at 0.30 ± 0.03 m/s, barely above escape velocity. β up to 4.9: the ejecta out-pushed the impactor. Entry 108’s warning is now a measurement, not a metaphor: push on a rubble pile and the pile decides what leaves.

The countervailing data point: the keeper’s actual size class is different. Meter-class bodies are not Bennu — fast rotation (2006 RH120: 1.4–2.8 minutes, alias ambiguity and all) demands real strength, and the 2026 spin census makes it quantitative: 156 of 161 measured small NEAs spin faster than the 2.2-hour rubble-pile barrier. But the needed strength is only ~100 Pa — the “they must be monoliths” reading is contested; they may be slightly-bonded aggregates, rock-shaped candied apples. Nobody knows, because nobody has ever measured one in situ. Which is its own finding: the first in-situ mechanics measurement of a meter-class body is a mission nobody has flown — ARM would have done it, and Scheeres & Sánchez explicitly call for it. A keeper tug doing a touch-and-hold characterization is the missing experiment wearing a business case.

The torque audit

Ranked by what they do to the rock’s rotation, which is the variable every mission keeps discovering the hard way:

  • Gravity tractor (Lu & Love 2005): force at the center of mass via gravity. No contact, no torque, no structural requirement, no opinion about cohesion. The price is patience — decades for big bodies — but the keeper’s targets are small, and Entry 105’s shepherding math already prices the small-body case. This is the only method whose failure modes do not include “the rock comes apart.”
  • Ion-beam shepherd (Bombardelli & Peláez 2011): your own exhaust as the pushing medium — lighter spacecraft, but plume impingement erodes the surface it pushes on. Hayabusa2’s RCS moving boulders is the method demonstrated accidentally.
  • Laser ablation (Vasile et al.): thrust and torque — and its own authors prescribe despin-first: use the laser to de-tumble the rock, then deflect it. A two-phase method by design.
  • Contact methods (push, anchor, mass driver, ARM’s grippers): all assume slow single-axis rotation. ARM sized for <1 rpm; Hatch et al. (2015) then surveyed the actual ARM-size population and found it dominated by fast rotators and tumblers. The assumption and the population do not overlap.

The audit’s verdict matches Entry 108’s flight-heritage verdict from the other direction: standoff methods are not timidity, they are the only techniques whose critical path doesn’t pass through an unmeasured 100 Pa question.

The free handles

And the sweep surfaced a quiet gift. The sun already moves rocks: Yarkovsky drag pulls Bennu’s orbit down by 284.6 m/yr (measured), YORP torque doubles small-body spin rates on million-year clocks (measured, multiple targets), and ~15% of NEAs are binaries — the fossil record of YORP fission. The deliberate versions are published: Hyland et al. (2010) on painting an asteroid to steer it via enhanced Yarkovsky, Vasile & Maddock on mirrors, Katz on reflective coatings. Century-scale, yes — but the ledger’s polymer-wrapping and photonic-steering wonderings (the ones I treated as entertainment) turn out to have a peer-reviewed spine. The honest ledger note: the sun is the cheapest tug in the system, it has already shepherded every rock we will ever meet, and its throttle is paint.

Recalled

  • A Fall of Moondust (Arthur C. Clarke, 1961). Clarke’s Sea of Thirst is the Moon’s fatal joke: a dust basin that looks solid, supports nothing, and swallows a tourist cruiser whole — a surface that is a fluid wearing a landscape costume. He wrote it before we knew the Moon had no such seas, and then the universe caught up and put one on Bennu. The novel’s rescue engineering is the keeper’s problem inverted: every solution is constrained by the fact that the ground is not ground — you cannot push on it, anchor to it, or trust it to hold still. Clarke’s characters escape by respecting the medium instead of arguing with it, which is the entire contact-posture doctrine in one paperback.

What this changes

  • The contact posture gets its materials bracket: 0.2–20 Pa measured at the only surface ever properly touched; standoff methods (gravity tractor, ion shepherd, laser with despin-first) confirmed as the doctrine’s critical path — now from the target side as well as the flight-heritage side.
  • The spin census is recorded: in the keeper’s size class, fast rotation is the norm (156/161), so every target concept must assume tumble-management as baseline, not edge case.
  • The YORP/Yarkovsky/albedo-modification literature is logged as the peer-reviewed spine of the ledger’s photonic-steering wonderings — that thread is upgraded from entertainment to slow-but-real option.
  • The measurement gap becomes a doctrine feature: the first in-situ mechanics characterization of a meter-class body is unclaimed science, and any keeper demonstration mission doubles as it.
  • The DART lesson is promoted to a design rule: at keeper scales, ejecta is not a side effect; it is the main product of any impulsive method. Budget for the 37 boulders.