Gregory Friedman The Greg Wide Web on the name →

Case Study · Physics · Kinematics

Three Worlds, One Parabola

2026

The idea

Everyone is taught that a thrown ball traces a parabola, and almost nobody is shown the one fact that makes the parabola interesting: its shape is set entirely by the strength of gravity, and by nothing else about the world it is thrown in. Change the planet and you change the arc; keep the launch the same and the arc is a fingerprint of gg. That is the claim this plate is built to make you see directly, by drawing the same launch on three worlds side by side.

The setup is the cleanest one physics has. Ignore the air (a real and large simplification I will come back to), and a projectile obeys two independent stories at once. Horizontally nothing pushes it, so it drifts at a constant speed. Vertically gravity pulls it down at a constant rate, so its height is a thrown-up, fall-back parabola in time. The horizontal drift stretches that vertical parabola sideways into the arc you see, and the two motions never talk to each other, which is the small miracle that makes the whole thing solvable with no calculus beyond completing a square.

The approach

Write the launch as a speed v0v_0 at an angle θ\theta above the ground. Split it into a horizontal part and a vertical part,

vx=v0cos⁡θ,vy=v0sin⁡θ,v_x = v_0\cos\theta, \qquad v_y = v_0\sin\theta,

and the position at time tt is just the two stories written down,

x(t)=vx t,y(t)=vy t−12g t2.x(t) = v_x\, t, \qquad y(t) = v_y\, t - \tfrac{1}{2} g\, t^{2}.

Everything you would want to know falls out of those two lines. The flight ends when yy returns to zero, at T=2vy/gT = 2v_y/g. The horizontal distance covered in that time is the range,

R=vxT=v02 sin⁡2θg,R = v_x T = \frac{v_0^{2}\,\sin 2\theta}{g},

which is why 45∘45^\circ throws farthest (it maximizes sin⁡2θ\sin 2\theta) and why a 30∘30^\circ and a 60∘60^\circ throw land in the same place. The peak height is reached at half the flight time, giving

H=vy22g.H = \frac{v_y^{2}}{2g}.

Notice where gg sits in all three results: on the bottom. Halve the gravity and you double the range and double the apex and lengthen the flight, all from the one change. The first plate computes these for Earth (g=9.81g = 9.81), the Moon (1.621.62), and Mars (3.723.72) and draws all three arcs together, faint for the two worlds you did not choose and bold for the one you did, so the effect of gravity is a gap between curves rather than a row in a table. The axes are in real metres at equal scale on both directions, so a 45∘45^\circ arc looks like a genuine 45∘45^\circ arc and not a squashed one.

45° at 20 m/s, by world

Range, apex, and flight time for a 45-degree 20 m/s shot on Earth, the Moon, and Mars.
worldrangeapexflight
Earth 40.8 m 10.2 m 2.9 s
Moon 246.9 m 61.7 m 17.5 s
Mars 107.5 m 26.9 m 7.6 s

Plate I A vacuum projectile on three worlds. Set the launch angle and speed; the bold green arc is the selected world, the faint arcs are the other two, and the ochre dots mark the apex and the landing while the short ochre stroke at the origin is the launch velocity. Both axes are real metres at equal scale, labelled with ticks, so the parabola keeps its true shape. Press Launch to send the marker along the selected arc (it holds at the apex under reduced-motion). The readout gives range, apex, flight time, and impact speed; switch worlds to watch the Moon's low gravity stretch the same throw far past its Earth landing.

The Moon arc is the one that makes the point. The same throw that carries about forty metres on Earth clears well over two hundred on the Moon, and it hangs in the air roughly six times as long, because the Moon’s gravity is about a sixth of ours and every one of range, apex, and flight time scales inversely with gg. Mars sits in between, a little more than a third of Earth’s pull. Three arcs from one launch is the whole argument that gravity, and only gravity, sets the shape.

What the air costs

Everything above is vacuum physics, and vacuum physics is an idealization rather than a rounding error. So the second plate charges for it. Put a projectile back in the air and the acceleration picks up a second term,

a=−g y^−km ∥v∥ v,k=12ρ CdA,\mathbf{a} = -g\,\hat{\mathbf{y}} - \frac{k}{m}\,\lVert\mathbf{v}\rVert\,\mathbf{v}, \qquad k = \tfrac{1}{2}\rho\,C_d A,

and the tidy structure of the first section is gone in one line. Drag points against the velocity, and the velocity has both components in it, so the horizontal story now depends on the vertical one and the vertical on the horizontal. The two independent motions become a coupled nonlinear system with no closed-form solution, which is why the plate below integrates rather than evaluates: fixed-step Runge-Kutta on the four-dimensional state (x,y,vx,vy)(x, y, v_x, v_y), with the landing interpolated onto y=0y = 0. The vacuum curve it is drawn against stays the exact formula from the first section, never a second integration, so any gap you see between the two arcs is physics and not integrator error.

Everything about the body reduces to one number, β=k/m=ρCdπd2/8m\beta = k/m = \rho C_d \pi d^2 / 8m, which is cross-section per unit mass with the air’s properties folded in. The three specimens are chosen to span it by three orders of magnitude: a table tennis ball (β≈0.134\beta \approx 0.134 per metre), a baseball (0.00850.0085), and a shot put (0.000380.00038). Launch all three at 45∘45^\circ and 2020 m/s, where the vacuum answer is 40.840.8 m for every one of them, and the air sorts them out: the shot put lands at 40.340.3 m and has lost one percent, the baseball at 32.432.4 m and has lost twenty-one, and the table tennis ball at 9.99.9 m, having lost seventy-six percent of its range to a fluid nobody drew. Nothing about the launch differed. Mass and frontal area did all of it.

The more interesting damage is to the symmetry. In vacuum the arc is its own mirror image: apex at half the range, impact speed equal to launch speed, descent as gentle as the climb. Against air every one of those fails, and fails harder the lighter the body. Impact speed drops from 2020 m/s to 19.819.8, 16.216.2, and 7.17.1 across the three. The apex slides past the midpoint, from 50%50\% of the range for the shot put to 52%52\% for the baseball and 62%62\% for the table tennis ball, which is the lopsided shape you actually see when someone hits a light ball hard: a long climb and a short steep fall. And the farthest-throwing angle walks down from 45∘45^\circ to 43∘43^\circ to 35∘35^\circ, because with drag in the picture there is a real penalty for buying height with time spent slowing down.

45° at 20 m/s · vacuum range 40.8 m

Range against air for three spheres launched at 45 degrees and 20 metres per second, against the vacuum range they would all share.
bodyrange in airrange lost
Table tennis 9.9 m 76%
Baseball 32.4 m 21%
Shot put 40.3 m 1%
Plate II One launch, drawn twice. The faint arc is the vacuum parabola of Plate I, computed in closed form; the bold arc is the same angle and speed integrated against air with quadratic drag on a sphere. Ochre dots mark the apex and the landing of the real arc. Switch bodies to sweep three orders of magnitude of drag: a table tennis ball, a baseball, and a shot put, all launched identically and all landing somewhere else. The cells below read the two arcs side by side, and the third one prices the difference: range lost, where the apex sits as a fraction of the range (exactly half only in vacuum), the angle that throws farthest in air against the vacuum's 45∘45^\circ, and the terminal speed g/β\sqrt{g/\beta} that decides which regime the launch is in.

The number that predicts which of those three regimes a launch is in is the terminal speed, vterm=g/βv_{\text{term}} = \sqrt{g/\beta}: the speed at which drag alone balances weight. It is 161161 m/s for the shot put, 3434 for the baseball, and 8.68.6 for the table tennis ball. Read the launch against it and the whole plate collapses into one ratio. The shot put leaves at a tenth of its terminal speed, so drag is a small correction to gravity and the vacuum parabola is nearly right. The table tennis ball leaves at more than twice its terminal speed, so for the first part of the flight drag is the dominant force and gravity is the correction. The baseball sits in between, which is exactly why a baseball is the interesting case and why it is the plate’s default.

The drag model has its own asterisks, and they are the same species as the vacuum one, just smaller. CdC_d is pinned at 0.470.47, the textbook value for a smooth sphere, when a real drag coefficient depends on the Reynolds number and drops sharply through the drag crisis as the boundary layer goes turbulent; a real baseball’s seams put it nearer 0.350.35 and are the reason a scuffed ball behaves differently from a new one. There is no spin here either, so no Magnus force, which means the drag plate can explain why a fastball falls short of its parabola but not why a curveball curves. Air density is constant, the air is still, and the ball is a sphere. What survives all of that is the shape of the effect and the ordering of the three bodies, which is what the figure claims; the individual metre is approximate.

What I would build next

Plate II is the asterisk on Plate I, and the useful consequence is that Plate I’s three worlds are wrong by three different amounts rather than by one. The Moon has no atmosphere worth the word, so the arc drawn for it is essentially exact. Mars has roughly a sixtieth of Earth’s surface air density, so its real trajectory sits a little inside the drawn one and the vacuum answer is a good approximation. Earth is the world the first figure actually misleads about, by about the twenty-one percent that the second one charges a baseball. Folding the drag correction into the three-world comparison itself, so that each world carries its own air rather than Plate II carrying Earth’s alone, is the join the two plates do not yet make, and it would also settle a question the pair currently raises and drops: whether a shot that clears two hundred metres on the Moon would clear anything like that on Mars.

Two smaller limits. The launch and landing heights are equal, so the symmetric formulas apply cleanly; a raised launch (a ball off a cliff, the true free-fall case at θ=90∘\theta = 90^\circ) needs the general quadratic for the landing time and is a natural next control. And the equations are all it draws, not the underlying vectors: a companion view that decomposes the velocity into its constant horizontal and its shrinking-then-growing vertical component, frame by frame, would show why the parabola is a parabola rather than just that it is one.

The idea, and the equations, come from some of the first real code I wrote, a small Java projectile-and-free-fall exercise; this is that exercise rebuilt in the browser, with the physics kept and the air added. It sits next to the site’s other “watch a rule unfold” figures: Newton’s method iterates toward a root the way this integrates toward the ground, and the slope-fields plate is the same kinematics written as the vector field these arcs are solutions of.

Keys on this page

A figure's keys are its own: they answer once the figure has focus, so tab to it or click it first. Esc closes this sheet.