Why galaxies spin too fast: add a dark halo and watch the rotation curve flatten.
The problem. A star on a circular orbit feels gravity from the mass M(<r) inside its orbit: v = √(G M(<r) / r). If almost all mass were the visible stars and gas, M(<r) would stop growing beyond the disc and v would fall as r-1/2 (Keplerian). Measured rotation curves of real galaxies stay roughly flat out to large radius, which is the classic evidence for an extra, unseen mass component.
Models used. Visible matter: a spherical exponential density ρ∝e-r/rd, with enclosed mass Mv[1 - (1 + x + x²/2)e-x], x = r/rd. Dark halo: the NFW profile ρ = ρ0/[(r/rs)(1+r/rs)²] whose enclosed mass is M200[ln(1+x) - x/(1+x)]/[ln(1+c) - c/(1+c)], with r200 from M200 = 200 ρcrit (4π/3) r200³ and rs = r200/c.
Stars. The 420 stars are real test particles integrated with a leapfrog scheme in the total gravitational field (0.15 kpc softening). Stars start on circular orbits for the current mass model. Setting the halo to zero lowers the whole rotation curve (Keplerian fall-off in the outskirts), which no longer matches the orange points.
Honest limits: real discs are flat, not spherical, so the visible curve is only an approximation (the exact flat-disc result differs by tens of percent near the peak); no gas, bulge or bar; the orange "observations" are synthetic illustrative points generated from a fiducial Milky-Way-like model with 4 % deterministic scatter, not real measurements; typical values only. Dark matter is one explanation; modified-gravity alternatives (MOND) also fit rotation curves but struggle with other evidence.