Einstein–de Haas — magnetism is real angular momentum

A rod hangs from a thin fibre inside a coil. Switch on the field and its atomic moments align — and the rod physically rotates. It must: a magnetic moment carries angular momentum, so aligning the moments (electronic L up) forces the body to spin the other way (conservation of total L). In RealQM this needs no spin primitive: a moment is a real circulating charge, so it carries real mechanical angular momentum by construction — the effect is native, not an extra postulate.
Side view. Arrows = atomic moments (green when aligned with B). The top pointer shows the rod's mechanical rotation — opposite to the electronic angular momentum.
magnetisation M 0.00
electronic L (moments) 0.00
body L (mechanical) 0.00
total L (conserved) 0.00
rod rotation 0.0°
inferred g = μ/L —
Slide B: the moments align, the rod counter-rotates. Historically the measured ratio gave g ≈ 2 — the magnetism is spin, not orbital (which would give g=1). The Barnett button runs the converse: spinning the rod magnetises it.