<!-- Generated by tools/athletics-doc.ts. Do not edit by hand. -->

Physical Tests

The same engine that resolves a gunnery shot resolves an arm-wrestle, a lift, and a jump. 3d6 exploding, roll-over, Margin drives the outcome. Nothing here is a special case.

Difficulties are derived from real biomechanics and real gravity, not invented.

The universal test

Roll 3d6, exploding. Add Skill and Attribute. Meet or beat the Target Number. Margin = Roll — Target Number.

Margin is read as degree of success: how far you cleared the gorge, how cleanly you pinned the arm, how much progress you made on the work. For extended tasks it accumulates against a threshold — see the Margin and Progress model.

Difficulty ladder

Anchored to the 3d6! distribution. An unskilled character rolls flat; skill shifts the whole curve.

Target Number 11 is the median and the anchor for every modifier in the game.

On the 21 threshold. In gunnery, Target Numbers past 21 are reported as "lock required" rather than rolled — that is a deliberate playability decision, and it should be owned as one rather than dressed up as physics. Real forces take sub-10% shots constantly, through suppression and volume of fire; the clamp exists because a single low-probability roll is bad table experience, not because the shot is impossible.

Skill tests are not clamped. A character may attempt an Extreme task at TN 24 and fail most of the time, because unlike a gunnery shot it is usually the only option available. The two rules are deliberately different and the difference is a design choice, not an oversight.

Opposed tests do not exist

C.A.T.S. has no opposed rolls. An arm-wrestling match is not two rolls compared; it is one player roll against a Target Number derived from the opponent.

Target Number = 8 + opponent's effective Strength rating

The GM never rolls. This keeps every die at the table in a player's hand, and it makes the Target Number a pure function of game state — which is what the Player Assistant needs to compute and display it.

For a sustained contest, use a progress threshold: first to accumulate the required Margin wins, with each roll representing a few seconds of straining.

Jumping, and why gravity is the interesting part

Projectile range for a given takeoff velocity:

range = v squared sin(2 theta) / g

Distance scales roughly as 1/g, the single most dramatic environmental effect in the game.

Not exactly 1/g, though. Takeoff velocity is not gravity-independent: the leg extends over a crouch of about 0.4 m producing roughly constant force, so the jumper accelerates at (a — g) and gets faster takeoff in low gravity. Low-g jumps are therefore somewhat better than a plain 1/g rule predicts, until muscle contraction speed becomes the binding limit rather than force.

Standing long jump, by takeoff velocity and gravity

Distances in metres, at a realistic 25-degree takeoff angle.

Measured toe-to-heel, so each figure includes the body-geometry term as well as the ballistic flight of the centre of mass.

Earth figures calibrate against real norms: untrained 2.0-2.2 m, trained 2.4-2.8 m, elite 3.0 m and up. A purely ballistic model lands about half of that, because a measured jump runs from toe line to heel strike while the projectile formula tracks only the centre of mass — worth about 1.1 m of limb geometry that does not scale with gravity.

A trained soldier clears about 1.2 m standing on Earth and about 7.5 m on Luna. That is not a bonus on a roll — it is a different tactical geometry. Gaps that are obstacles on Earth are irrelevant on Ceres, and a powered-armour trooper on Luna crosses a street.

Where this model stops being a jump. On Ceres the figures pass 100 m, and at that point the flight lasts the better part of a minute. That is not a jump, it is a ballistic hop: committed at takeoff, unsteerable without thrust, and landing at takeoff speed. Treat anything over roughly 20 m as its own manoeuvre with its own risks — you cannot dodge mid-flight, you are a predictable target for the whole arc, and the landing needs its own test. The physics is right; the word "jump" stops being.

Vertical jump

Vertical jump is measured as centre-of-mass rise, so it needs no geometry term — which is precisely why the vertical numbers were right when the long-jump numbers were not.

Ceiling height becomes a hazard, not a convenience. A trooper who jumps in a 2.4 m corridor on Luna hits the ceiling head-first at speed. Low gravity is dangerous long before it is useful, and that is a rules consequence worth keeping.

Lifting

Muscle produces roughly the same force regardless of local gravity, so liftable mass scales as 1/g. Deadlift capacity for an 80 kg human:

The inertia trap

This is the physics that science fiction gaming almost always gets wrong, and it is worth a rule of its own.

Gravity changes weight. It never changes mass. A crate that is easy to hold in low gravity is exactly as hard to stop as it was on Earth.

A 400 kg crate on Luna weighs about 66 kgf — a trooper can lift it one-handed. Getting it moving at walking pace and then stopping it takes 800 N·s, the same as on Earth, which is roughly the impulse of catching a falling motorcycle.

Rule consequence: tests to lift or carry use weight and scale with gravity. Tests to start, stop, catch, or change the direction of a mass use inertia and ignore gravity entirely. Low-gravity cargo handling is dangerous precisely because the first test gets easier while the second does not.

Applying these to the table

Every number here is provisional until playtested.