Finding the same frequency tomorrow in a thorium clock

Stability over a day

The thorium-229 clock’s laser can follow a nuclear transition for a day. The measurement itself is the news: nuclei inside a crystal anchor the feedback loop that corrects the laser frequency. Comparing it with an atomic clock and averaging over a day brings instability toward 10^-15. That is a useful result. The instructive part is that this statistic does not describe a measurement established again later. How quietly the clock runs and whether tomorrow’s run starts at the same frequency require different tests.^[1]

The apparatus makes the distinction tangible. A laser beam approximately 0.5 mm across illuminates only part of a crystal 3.1 mm across. Realigning the laser can change that region. Frequency reproducibility between days is approximately 5 × 10^-13. Long averaging reduces random fluctuations during a run; it does not automatically correct a different crystal region giving a different line centre. I therefore value a performance description that includes both within-run instability and changes after realignment.^[1]

Where in the crystal?

The researchers tested this with five scans at four positions. One position was measured twice. Each scan collected 20-second measurements over 20 minutes. Line-centre differences reached 1.7 kHz across positions, compared with 0.13 kHz expected from statistical spread alone. Agreement between the two scans at the same position gives a concrete reason to track where the beam lands. A small number of positions does not map the entire crystal, but it exposes an additional variable introduced when the apparatus is set up again.^[1]

Local mechanical strain is one explanation proposed by the authors. It has not been established as the sole cause of the frequency differences. Other apparatus conditions can also change during realignment. I read this comparison as a robustness test identifying a condition to control, rather than an experiment settling the cause. Revisiting the same position, comparing other regions and documenting conditions that change with alignment make the reproducibility claim more precise. Giving the result its due does not require calling this limitation a failure.^[1]

The search timescale

The distinction matters still more for the dark-matter search. The experiment looks for oscillations in the frequency ratio between nuclear and atomic clocks, a possible effect of hypothetical fields. No significant signal was found. Selecting the largest fluctuation among approximately 3800 independent frequencies requires a higher threshold than assessing a fluctuation at one frequency. The team accounted for that multiple-search effect with 1000 simulations. This is a substantive methodological strength aimed at preventing a chance peak in a null search from resembling a discovery.^[1]

A measurement seeking fast oscillations carries different error sources from one seeking slow frequency drift across days. Position-dependent line-centre shifts particularly constrain the second question. My criterion for this clock extends beyond the smallest instability figure: the frequency comparison must track the same physical conditions over the timescale of the claim. The study advances the field by showing both working feedback and the boundary of that condition. Stronger long-term interpretation rests on measuring and controlling alignment and position changes within the frequency-error budget.^[1]