Displace the magnet by a small angle and the restoring torque mB sinθ ≈ mBθ makes it swing like a pendulum, with period T = 2π√I/mB — I its moment of inertia. Time the swings and you have a balance for magnetic quantities: comparing periods gives the ratio of two fields (or two moments) without ever measuring B directly. Heavier or weaker → slower; stronger field → faster.

A uniform field pulls equally on both poles, so there's no net force — only a torque τ = mB sinθ spinning the magnet toward alignment. Doing that against the torque stores energy U = −mB cosθ: lowest (stable) pointing along the field, highest (unstable) pointing against it, and the torque peaks broadside at θ = 90°. Swing the angle and read both.

Far from a short dipole the field is B = (μ₀/4π)(m/d³)√1+3cos²θ, where θ is measured from the axis. Straight off the ends (axial, θ=0) it is twice as strong as straight out to the side (equatorial, θ=90°), and it always tips toward the axis by tanα = ½ tanθ. Drag the point P around and watch B swing.

Give each end a pole strength q (in A·m) a distance 2ℓ apart, and the magnet carries a magnetic moment m = q·2ℓ, an arrow along S→N. Field lines run N → S outside and unbroken S → N inside — always closed loops, so there is never a source or a sink. That is why you cannot isolate a pole: snap the magnet and each fragment instantly grows its own N and S.