Normalize spin noise by the signal it preserves

Use dimensionless collective operators $J_k=\frac12\sum_j\sigma_k^{(j)}$ for N spin-one-half particles. A coherent-spin state polarized along x has mean $\langle J_x\rangle=N/2$, zero mean transverse components, and transverse variance $N/4$. Physical angular momentum would multiply J by $\hbar$, but using dimensionless spins keeps the normalization explicit. The uncertainty relation between transverse components then constrains how much one noise direction can be compressed while maintaining a large mean spin.

A small phase rotation about y moves the mean population difference $J_z$ with slope whose magnitude is $|\langle J_x\rangle|$ at the operating point. Linear error propagation for that readout divides the standard deviation of $J_z$ by this slope. For the coherent-spin reference the result is $1/\sqrt N$. This describes a small-angle calibrated estimator using collective population readout; it is not a claim about every possible measurement on the state or about an arbitrary large-angle likelihood.

The Wineland parameter $\xi_R^2$ is the phase variance inferred by this readout divided by the coherent-spin phase variance at the same N. Define contrast $C=2|\langle J_x\rangle|/N$ and normalized transverse variance $v=4\operatorname{Var}(J_z)/N$. Then $\xi_R^2=v/C^2$. A state with half the reference variance has $v=1/2$, but it gains nothing if its remaining contrast is too small. Signal loss enters quadratically, so comparing only histograms of population noise can give the wrong experimental conclusion.

When the parameter is less than one, this readout offers sub-shot-noise small-angle sensitivity under the stated calibration. The reciprocal is a phase-variance gain; its square root is the standard-deviation gain. Squeezing is often reported in decibels, which use ten times the base-ten logarithm of the variance ratio. A fourfold variance gain is about six decibels and a twofold standard-deviation gain. State whether a plotted number is a ratio, its square root, or a logarithmic representation before comparing experiments.

Squeezing transfers uncertainty into another collective direction rather than removing the uncertainty relation. Imperfect pulse phases can rotate an enlarged noise direction into the measured component, especially when the anti-squeezing is large. Preparation may also reduce spin length, scatter particles, or take extra time. Detection noise adds to the measured population variance unless it is separately characterized. A gain quoted after subtracting detector noise describes a different operational quantity from the precision achieved by the complete readout chain.

For spin-one-half ensembles, metrological squeezing below one witnesses useful particle correlations under the stated particle-count and collective-spin assumptions. The converse is false: some useful non-Gaussian entangled states have little mean spin and are poorly described by this particular ratio. Quantum Fisher information can recognize advantages available through other measurements. Treat the squeezing parameter as a direct diagnostic for a specified collective readout, and retain the full likelihood or a broader information analysis when the mean-spin linear calibration becomes inadequate.