Light prepares a collective orientation
A tilted spinning top traces a slow circle around the vertical. An atomic magnetic moment in a field undergoes a related motion: its direction precesses around the field axis. This is Larmor precession. A stronger field produces faster precession. The instrument reads the combined response of many atoms.
Initially, the atomic moments point in many directions and largely cancel. Circularly polarized light carries angular momentum along its propagation direction. Repeated absorption and emission give the atoms a preferred orientation, a process called optical pumping. Collisions and the cell walls gradually erase it, so continuous pumping replenishes the orientation.
A second, weak, linearly polarized beam can probe that orientation. Its polarization rotates slightly as it passes through the vapor. The angle depends on the spin projection along the probe. Atomic motion becomes changing optical polarization, which photodetectors convert into a time-dependent signal.
A resonance supplies the frequency ruler
The Mx method sustains a readable precession with a transverse oscillating magnetic field, much as repeated pushes sustain a swing. The conventional name is RF drive even when its frequency lies in the audio range. The optical response changes strongly as the drive approaches the atomic precession frequency.
Sweep the drive through that resonance and measure amplitude and phase at each frequency. Phase tells which oscillation leads or lags. A lock-in detector compares the light signal with a reference and a quarter-cycle companion. The two signed slow outputs retain information that an intensity peak alone would lose.
At low field, a selected cesium ground state supplies a known scale: roughly three and a half extra cycles per second for each extra nanotesla. The fitted resonance centre therefore becomes a field magnitude. Brighter probe light increases the counts; it does not directly stretch this frequency ruler. The frequency counter still needs a reliable calibration.
The response also has a width. Relaxation, pumping and RF drive affect it, and stronger drive can broaden it. A sweep needs points on both sides to distinguish the centre from the width. Unfavourable pump, drive or probe directions can also weaken the signal until the precession becomes invisible.
Probe drift and pump drift affect different things
A polarizing splitter sends the probe to two detectors. A small rotation gives one detector more light and the other less. Dark observations set the zero level; known rotations calibrate the two gains. The corrected difference divided by the corrected sum then measures polarization.
If the probe grows brighter, both channels acquire nearly the same multiplicative factor. Their normalized difference cancels that common factor. An independent brightness monitor checks the change in source power. Saturation and separate detector-gain drift still need attention, so the optical standards remain part of the measurement.
Stronger pumping changes the atoms themselves. It changes their orientation, their return to steady state and the resonance width. A normalized probe ratio retains those atomic changes. Pump drift during a sweep can distort a stationary resonance curve, which is why a separate pump monitor must accompany the optical record.
Dark observations, known optical rotations, oscillator references and the two sides of a resonance each constrain a different part of the measurement. Keeping them together makes a frequency shift easier to trace to its physical cause.
Near zero field, read a static tilt
A sufficiently weak field permits another measurement. Continuous pumping orients the spins, a small transverse field tilts them slightly, and the probe reads that static tilt. RF can be switched off. A known coil sweep then reveals an optical response that changes sign near zero field.
Its slope depends on pumping and relaxation as well as on field. Farther from zero, the response grows and then falls, so two fields can give the same rotation. Independent relaxation information and a local calibration define a monotonic interval where a conditional field inference is possible.
Resonant precession and static reorientation use the same atoms under different operating conditions. Extending a fast-precession approximation all the way to zero loses the basis of that approximation. Choosing the appropriate response, with optical standards and timing references retained, turns the atoms into an interpretable magnetic ruler.