Wuhan–Shanghai optical clock comparison hits SI-second benchmark, preprint reports
Researchers in China say they have carried out the first intercity optical-clock comparison over a long-haul fiber network at a precision level that meets a key international benchmark for a future redefinition of the SI second. In an arXiv preprint posted this month, the team reported a Wuhan-Shanghai comparison over 1,350 kilometers, or about 839 miles, of telecom fiber with a total fractional standard uncertainty of 2.08 × 10^-18.
The preprint, “Yb/Sr optical clock comparison at 2.08E-18 uncertainty over a 1350-km fiber link,” is not yet peer-reviewed. The arXiv record lists an initial version posted Oct. 25 and a revised version on Nov. 14; the PDF is dated Tuesday.
The work involved a ytterbium-171 optical lattice clock in Wuhan and a strontium-87 optical lattice clock in Shanghai. The authors are affiliated with the University of Science and Technology of China and Hefei National Laboratory, the Innovation Academy for Precision Measurement Science and Technology of the Chinese Academy of Sciences in Wuhan, and Wuhan University. The two clocks were linked through a phase-stabilized 1,350-km fiber connection running over a commercial telecommunications network.
The team reported a Yb/Sr frequency ratio of 1.2075070393433377168(25). For nonexperts, the more important figure is the uncertainty: 2.08 × 10^-18, with a Bayesian statistical uncertainty of 0.94 × 10^-18. The authors say that is, to their knowledge, the first intercity long-distance frequency-ratio measurement to reach the uncertainty level called for in the international roadmap for validating a future redefinition of the second.
That matters because optical clocks are the leading candidates to eventually replace the current cesium-based definition of the SI second. The metrology roadmap cited in the paper says independent cross-laboratory frequency-ratio measurements need overall uncertainty below 5 × 10^-18 before a redefinition can move forward. A few comparisons at that level have been reported before, but mainly between clocks in the same place rather than clocks in different cities connected by a long-haul link.
The new result came from a campaign spanning 58 calendar days in 2026, with 274 effective hours of synchronous operation. The final value combined 16 jump-free data segments. According to the uncertainty budget in the paper, the largest contributions came from the clocks themselves and from correcting for the two sites’ difference in gravitational potential, not from the fiber connection. The ytterbium clock systematic uncertainty was 1.3 × 10^-18, the strontium clock contribution was 0.92 × 10^-18, and the static geopotential correction contributed 0.946 × 10^-18. By contrast, the fiber link and the frequency comb each contributed less than 0.1 × 10^-18.
That gravity correction is not a side issue at this level of precision. Because gravity changes how fast clocks tick, clocks in different places must be corrected for their height and gravitational-potential difference. The paper gives a static geopotential difference between the sites of 280.042 ± 0.085 square meters per square second, based on leveling and checked against an estimate using satellite positioning and a geoid model. As a rule of thumb, about 1 × 10^-18 in clock frequency corresponds to roughly 1 centimeter, or 0.4 inches, of height.
The same sensitivity also let the researchers see a geophysical signal in the comparison data. The paper says the measured frequency ratio showed a significant correlation with Earth tides — small, periodic changes in Earth’s shape and gravitational potential caused mainly by the moon and sun. The measured response, the authors wrote, agreed with the model within the statistical uncertainty, pointing to the potential for long-distance optical-clock networks to support chronometric leveling, a way of inferring gravitational potential from clock rates.
Still, the result is one data point in a field that is still being cross-checked at the highest levels of precision. The paper notes that its ratio differs from a 2026 NIST-JILA Yb/Sr value by -5.16 × 10^-18, about 1.4 times the combined standard uncertainty. The authors present that not as a contradiction but as a reminder that more independent measurements are needed. As the paper quotes from the 28th General Conference on Weights and Measures, or CGPM, “there are noticeable inconsistencies in some comparisons of optical frequency standards and in the global set of frequency ratio measurements involving OFS.”