Testing granularity
Use established precision instruments to search for a predicted deviation or bound its size. This is a metrology and fundamental-physics experiment; the existing clock remains the reference instrument.
07 / THE EXPERIMENTAL BRIDGE
Make the prediction
survive the laboratory.
The first success is a well-posed, reproducible test—whether it reveals a signal or sets a useful limit.
Define the observable and the null model
Specify the measured fringe contrast or phase correlation, the proposed extra term, and the conventional model against which it is tested. Preregister a primary outcome and the conditions that would rule out the selected parameter range.
Design a reference-aware comparison
Choose transitions, interrogation sequences, and a reference architecture. Shared-laser measurements may suppress certain common noise contributions but can also suppress a common hypothesized signal. Different species require a real optical transfer architecture, not an assumed perfect shared reference.
Measure and deliberately vary the backgrounds
Change interrogation time, optical power, magnetic conditions, temperature, and synchronization conditions where appropriate. Record auxiliary channels and instrument calibration. Check whether the proposed signal follows controlled technical changes.
Fit uncertainty, not just a smooth curve
Estimate nuisance parameters and correlations; test model identifiability; quantify confidence intervals and systematic uncertainties. Use simulated signal injection to measure detection performance and false-positive behaviour before interpreting a small residual.
Replicate with independent instruments
Repeat the protocol with another team or architecture. Agreement across laboratories is particularly valuable when the claim concerns a universal process. Publish methods, calibration records, analysis code, and the conditions under which a null result is informative.
Building a better clock
Identify an accessible physical signal, a coupling and readout mechanism, a stable reference, and an estimator. Demonstrate better performance under comparable conditions. This engineering objective goes beyond observing—or constraining—decoherence.