{"id":"240539fd-d909-4c4f-ad16-b7f66ab672cc","arxiv_id":"2508.13140","paper_version":1,"verdict":"UNVERDICTED","confidence":"LOW","novelty_score":6.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":0,"one_line_summary":"A rubidium atomic fountain at USNO shows TDEV of 8 ns at about 3 years and a holdover of the BIPM timescale within 14 ns over 12 years of continuous data.","lead":"Four rubidium atomic fountains at the U.S. Naval Observatory have run for 12 years as part of the global atomic timescale. The best fountain keeps time to within 14 nanoseconds of the world's best timescale over that whole period.","discovery_kind":"extension","skeptic_critique":{"model":"deepseek-v4-flash","headline":"EAL self-reference could inflate fountain holdover; independence test required","rationale":"The reader correctly identified data completeness and the fountains' EAL contribution as weak points, but placed primary weight on uninterrupted data. My analysis elevates the EAL self-reference to the single most load-bearing concern because it can bias the headline metric even if the data are perfectly continuous. The reader's verdict of UNVERDICTED is appropriate, and my concern does not change that verdict; it reinforces the need for the full paper and an independence analysis. The proposed concrete test would settle whether the self-reference actually matters: if removing the fountains from EAL leaves the holdover essentially unchanged, the central claim stands; if not, the claim is inflated. Since this is an abstract-only review, no definitive rejection or acceptance is justified, hence UNCHANGED rather than a more decisive verdict.","tokens_in":718,"tokens_out":2587,"duration_ms":27383,"concrete_test":"Using BIPM published EAL weight tables and clock comparison files for MJD 56074-60429, compute the monthly total weight of the four USNO Rb fountains in EAL. Then reconstruct the BIPM best timescale (or EAL) with all four fountains' data removed, and recompute the highest-performing fountain's TDEV at ~3 years and its 12-year holdover against this fountain-free reference. If the holdover degrades by more than ~20% (e.g., from ±14 ns to >±17 ns), the self-reference contamination is significant; if it remains within ~20%, the claim is robust. As a secondary check, confirm that the 12-year comparison record has no missing segments by cross-referencing the BIPM Circular T files.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The central claim is that a USNO rubidium fountain achieved TDEV of 8 ns at ~3 years and holdover of BIPM's best timescale of ±14 ns over 12 years. The most load-bearing concern is circularity: the abstract states these four fountains are the first cold-atom clocks to contribute to EAL, so they are inside the very ensemble from which BIPM's best timescale is derived. If the fountains carry non-negligible weight in EAL, then comparing a fountain to BIPM's best timescale is partially a self-comparison. The fountain's own noise is smoothed by the ensemble average, and the apparent holdover could be dominated by the ensemble's stability rather than by the fountain's independent performance. This would inflate the reported TDEV and holdover in a way that a reader could mistake for a property of the fountain alone. The abstract does not quantify the fountains' total weight in EAL, nor does it provide an independent reference. A separate but related concern is data completeness: '12 years of uninterrupted data' must be verified against BIPM clock comparison files, since undocumented gaps could bias TDEV. However, the self-reference issue is more fundamental because it affects even a complete dataset. Without an exclusion or weight analysis, the 100-ns-level holdover claim is not established.","agreement_with_reader":"partial"},"referee_report":{"model":"deepseek-v4-flash","summary":"The manuscript reports on the long-term timing performance of four rubidium atomic fountains operated at the U.S. Naval Observatory, which the authors identify as the first cold-atom clocks to contribute to the free-running international atomic timescale EAL. Based on 12 years of BIPM comparison data (MJD 56074 to MJD 60429), the abstract claims that the highest-performing fountain achieves a TDEV of 8 ns at roughly 3 years and a holdover of BIPM's best timescale of ±14 ns at 12 years. The central claim is that a continuously operating rubidium fountain can serve as a 100-ns-level timing reference for over a decade.","tokens_in":955,"tokens_out":1744,"duration_ms":18166,"significance":"If the reported performance is substantiated, this would be an important demonstration: a cold-atom clock operating continuously in an operational environment, maintaining timing at the 100-ns level over 12 years. The use of a 12-year BIPM record is a strength, as is the concrete reporting of TDEV and holdover values. However, the significance is conditional on the comparison being meaningfully external to the clock under test and on the integrity of the reported uninterrupted data. The abstract alone does not establish these conditions, so the significance cannot yet be fully assessed.","major_comments":[{"comment":"The holdover comparison is not fully external: the four fountains are stated to contribute to EAL, from which BIPM's best timescale is derived. Unless the abstract or full text quantifies the fountains' total weight in EAL, or provides an analysis after excluding them from the ensemble, the reported TDEV and holdover may partly reflect the ensemble averaging of the timescale rather than the independent stability of the fountain. This is a load-bearing point for the 12-year holdover claim and should be addressed explicitly.","section":"Abstract"},{"comment":"The phrase '12 years of uninterrupted data' is a load-bearing premise, but the abstract gives no information about data gaps, operational interruptions, or any exclusions applied to the BIPM comparison record. TDEV estimates are sensitive to missing or excluded segments, and the paper should state the completeness criteria, the handling of any gaps, and the provenance of the 'uninterrupted' designation.","section":"Abstract"},{"comment":"The reported values (TDEV of 8 ns at ~3 years, holdover of ±14 ns at 12 years) are presented without uncertainty intervals or confidence statements. For a claim at the 100-ns level, the comparison uncertainty, including the uncertainty of the BIPM link and the timescale itself, must be reported so that the reader can judge whether the quoted values are statistically distinguishable from a null result.","section":"Abstract"}],"minor_comments":[{"comment":"The abstract does not define TDEV or the exact meaning of 'holdover of BIPM's best timescale of ±14 ns'—in particular, whether the ±14 ns is a peak deviation, a standard deviation, or another statistic.","section":"Abstract"},{"comment":"The claim that these are 'the first cold-atom clocks to contribute to EAL' is a useful novelty statement, but it would be clearer to specify the exact dates of the contribution and the weight they receive in EAL.","section":"Abstract"},{"comment":"The notation 'MJD 56074 to MJD 60429' is precise, but the abstract would benefit from stating the corresponding calendar years for readers not familiar with modified Julian dates.","section":"Abstract"}],"recommendation":"major_revision","confidential_remarks":"This is an abstract-only submission, so the referee report is necessarily limited. The central claim is plausible but hinges on the self-reference issue with EAL and on the completeness of the 12-year dataset. If the full paper contains a weight analysis or exclusion test, it should be brought to the fore; without such an analysis, the claim as stated is not fully established. The paper may be a good fit for the journal if the full text resolves these points."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Here's my take on the fountain holdover paper.\n\nThe headline result is concrete and new: 12 years of continuous operation of four rubidium fountains, with the best one showing 8 ns TDEV at ~3 years and ±14 ns holdover versus BIPM's best timescale. That is exactly the kind of long-baseline evidence the timing community has been missing, and the authors are in a unique position to provide it—USNO has been running these clocks operationally since 2011.\n\nWhat the paper does well, based on the abstract, is state the measurement conditions clearly: the data span is given in MJD, the clocks are identified as the first cold-atom contributors to EAL, and the metric is TDEV rather than a more forgiving Allan deviation. That is a good sign. The novelty is the baseline, not the physics.\n\nThe soft spots are real but manageable. The first is the one the stress-test flags: the fountains are inside EAL, the very ensemble from which 'BIPM's best timescale' is constructed. The abstract does not quantify the fountains' weight. In practice, EAL is a weighted average of many clocks, so four rubidium fountains are likely a small component, but 'likely' is not good enough. The full paper needs to report the fountains' total weight in EAL over the 12-year interval, or better, compare against an EAL that explicitly excludes them. Without that, the holdover number is not fully independent.\n\nThe second issue is data completeness. '12 years of uninterrupted data' is a strong claim. BIPM comparison files are messy; there are gaps, outliers, and operational exclusions. TDEV is sensitive to how those are handled. The abstract gives no uncertainty intervals and no discussion of exclusions. This is a normal gap between an abstract and a metrology paper, not a fatal flaw.\n\nOne wording quibble: 'holdover of BIPM's best timescale' is a loose use of 'holdover,' which usually means timekeeping after loss of a reference. The comparison here is a long-term frequency/phase evaluation against an external timescale. The meaning is clear, but a referee should ask for precise language.\n\nWho is this for? The timing and metrology community. If the EAL-weight analysis and data-quality details are in the paper, this is a solid operational result that could influence how cold-atom fountains are used in timing infrastructures.\n\nRecommendation: send it to peer review. The claim is important enough and the group credible enough that a serious referee, focused on the independence and completeness questions, is warranted. If the paper lacks those details, it needs revision; if it has them, it is publishable. Either way, it should not be desk-rejected.","headline":"A credible long-baseline fountain holdover claim that deserves a serious referee, provided the paper quantifies the fountains' weight in EAL and verifies data continuity.","tokens_in":1448,"tokens_out":3052,"would_cite":true,"duration_ms":30537,"reading_group":"maybe","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":[],"pacs":[],"model":"deepseek-v4-flash","headline":"Rubidium atomic fountains, operated continuously since 2011, hold time to within ±14 nanoseconds of the best international timescale over 12 years.","keywords":["rubidium atomic fountain","cold-atom clock","continuous clock operation","time deviation (TDEV)","holdover","international timescale","long-term timing stability"],"falsifier":"A day-by-day reanalysis of the published comparison data for the 12-year interval that finds any missing segments or data flags would invalidate the uninterrupted-data assumption; additionally, recomputing TDEV and holdover from the same data should reproduce 8 ns and ±14 ns, and any large discrepancy would settle the claim.","tokens_in":539,"feed_emoji":"⏱️","tokens_out":5976,"duration_ms":54558,"temperature":0.7,"pith_summary":"This paper reports the first multi-year timing record from cold-atom clocks running continuously as timekeepers, not just as intermittent frequency standards. Four rubidium fountains at a national observatory have operated without interruption since 2011, and their comparison data over 12 years show that the best fountain drifts no more than ±14 nanoseconds from the international timescale. If the record holds, a single cold-atom fountain can replace the role of caesium beams and hydrogen masers in maintaining 100-nanosecond-level timing over a decade.","feed_headline":"Rubidium fountains hold time to ±14 ns for 12 years","feed_subtitle":"Continuously running cold-atom clocks pass a 12-year timing test against the international timescale.","key_machinery":"The load-bearing devices are rubidium atomic fountains: laser-cooled rubidium atoms are launched upward, pass through a microwave cavity, return under gravity, and their Ramsey interference fringes lock a local oscillator to the atomic transition. Their enabling feature for timing is continuous operation: four fountains have been running without interruption since 2011, producing a 12-year dataset that lets the paper evaluate long-term stability via time deviation (TDEV) and holdover against the international timescale.","core_discovery":"The central discovery is that rubidium atomic fountains can serve as continuously operating timing references with decade-long holdover. On the basis of 12 years of uninterrupted data from the international timekeeping bureau, the highest-performing fountain shows a time deviation of 8 nanoseconds at roughly 3 years and stays within ±14 nanoseconds of the best international timescale for the full 12 years. These fountains are also the first cold-atom clocks to contribute directly to the free-running international atomic timescale, which until now has been steered by conventional clocks.","pith_inferences":["The reported holdover is measured against the international timescale, which the fountains themselves help steer; a fully independent estimate of their free-running performance would require a comparison with an ensemble that excludes them.","If the TDEV continues to fall with averaging time beyond three years, the limiting noise source is not yet identified; if it flattens, environmental or flicker effects dominate, and a shorter record would have missed that.","The same continuous-operation strategy could be tested on optical clock transitions, for which the technology is closer to round-the-clock running than it was for fountains in 2011.","A reanalysis splitting the 12-year record into consecutive intervals would reveal whether the reported stability is stationary or depends on the particular period."],"forward_implications":["National timing systems could use a single cold-atom fountain as their reference and keep 100-nanosecond-level accuracy for a decade.","Cold-atom clocks have moved from research devices to operational timekeepers, closing the usual technology lag for timing applications.","The 12-year record provides a concrete stability target for future continuous clocks: an 8 ns time deviation at 3 years and ±14 ns holdover at 12 years.","The fact that these fountains contribute to the international timescale means cold-atom clocks can participate in ensemble timescales without degrading them."],"supporting_citations":[],"fun_headline_variants":["Atomic fountains hold time to 14 ns for 12 years","Rubidium fountains keep world time within 14 ns for 12 years","12-year timing test passed by rubidium fountains with 14 ns precision","Cold-atom clocks keep time accurately for 12 years: ±14 ns","Fountains hold time to ±14 ns for over a decade"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The 12-year record is truly uninterrupted and complete, so no unrecorded gaps or degraded data are hiding inside the reported numbers.","fun_headline_variants_meta":{"raw":{"variants":["Atomic fountains hold time to 14 ns for 12 years","Rubidium fountains keep world time within 14 ns for 12 years","12-year timing test passed by rubidium fountains with 14 ns precision","Cold-atom clocks keep time accurately for 12 years: ±14 ns","Fountains hold time to ±14 ns for over a decade"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000357,"raw_usage":{"total_tokens":1885,"prompt_tokens":843,"completion_tokens":1042,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":459,"completion_tokens_details":{"reasoning_tokens":946}},"tokens_in":459,"tokens_out":1042,"duration_ms":9805,"temperature":1.0,"reasoning_tokens":946,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-15T17:14:30.162103+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"A day-by-day reanalysis of the published comparison data for the 12-year interval that finds any missing segments or data flags would invalidate the uninterrupted-data assumption; additionally, recomputing TDEV and holdover from the same data should reproduce 8 ns and ±14 ns, and any large discrepancy would settle the claim.","supporting_citations":[],"review_version":2}