{"id":"2133c0fd-eca7-4a9f-83d7-74bd3a6a3e07","arxiv_id":"2505.05260","paper_version":1,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":6.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":5,"one_line_summary":"The 479-day radial velocity signal once attributed to a 3.9 Jupiter-mass planet around 42 Dra decays in amplitude and appears in COBE/DIRBE photometry, ruling out a planetary companion.","lead":"Long-term radial velocity and infrared photometry show that the previously claimed giant planet around the K giant star 42 Draconis is not real. The signal changes amplitude and appears in starlight itself, pointing to stellar oscillations rather than an orbiting companion.","discovery_kind":"replication","skeptic_critique":{"model":"deepseek-v4-flash","headline":"The 'conclusive' DIRBE photometric refutation in §4.4 is not established: the COBE/DIRBE 1.25 µm time series covers only ~10 months, shorter than the 479 d signal, so the low-frequency peak may be an unresolved trend.","rationale":"The reader's weakest assumption is on point: the DIRBE time baseline is never reported, and the cryogenic mission length makes a 479 d period unconfirmed. I agree with the CONDITIONAL verdict. The paper's central negative result is independently supported by the RV measurements: the 479 d signal's K-amplitude falls from roughly 110 m/s to 27 m/s, and the periodogram of the 2008–2011 data shows no peak at the orbital period even though a simulation shows it should be detected. That alone contradicts the single-companion interpretation. However, the manuscript repeatedly uses the DIRBE photometry as the 'conclusive' and 'firm' evidence that the variations are intrinsic stellar variability (oscillatory convection modes). If the DIRBE baseline is about 300 days, the apparent 479 d periodicity could be a low-frequency trend; the bootstrap FAP is computed for white noise at a specified frequency and would not capture this. The correct remedy is not rejection but revision: report the time baseline and sampling, test the peak after detrending, and soften the abstract/conclusion language. Therefore the reader's CONDITIONAL verdict should stand unchanged.","tokens_in":14426,"tokens_out":9643,"duration_ms":108804,"concrete_test":"Obtain the actual DIRBE timestamps for 42 Dra from Price et al. (2010) or the COBE archive; if the total span T is shorter than 479 d, recompute the periodogram after subtracting a linear trend, and compare the observed power at ν = 0.0021 d^-1 with the distribution of maximum low-frequency power obtained from a bootstrap that generates linear-trend-plus-noise light curves with the same timestamps. If the observed peak is not significant (FAP > 0.01) or its frequency tracks the lowest resolved frequency, then the Section 4.4 claim that DIRBE photometry rules out the planet is not supported. The paper should also state T and the number of independent epochs explicitly.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The abstract and Section 4.4 rest the 'conclusive' no-planet claim on the DIRBE 1.25 µm periodogram: a peak at ν = 0.0021 d^-1 (P = 479 d) with bootstrap FAP ≈ 2 × 10^-3, plus a 169.8 d peak. However, Section 4.4 does not state the time baseline of the DIRBE photometry. The Price et al. (2010) data are weekly averages from the cryogenic COBE mission, which lasted about 10 months (~300 d) for the 1.25 µm channel. A 479 d signal is therefore longer than the entire time series; at frequencies below the Fourier resolution (1/T ≈ 0.0033 d^-1), a monotonic or low-order trend produces broad low-frequency periodogram power, and the 'peak at the orbital frequency' is not resolved. Phasing the data in Figure 7 covers only about 0.6 of one cycle, so a sine fit is not distinguishable from a line. The windowed bootstrap at a known frequency tests white noise at that frequency, but not the null of a low-frequency trend, and no detrending is reported. Thus the photometric evidence for intrinsic stellar origin at 479 d—and the wording 'conclusively ruled out' / 'firmly establish'—is not secure. The independent RV evidence (amplitude drop from 96.6 to 27.4 m/s and absence of the 479 d period in the 2008–2011 subset) remains strong, so the underlying refutation of the published 479 d companion can survive; the manuscript should report the DIRBE baseline and downgrade the photometric claim to supporting rather than conclusive.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"The manuscript reports new radial velocity (RV) measurements of the K giant 42 Dra spanning 2004–2018 and re-examines the previously claimed 479-day, 3.9 Jupiter-mass companion. The authors find that the 479-day signal is not stable: the RV amplitude declines from ~96 m/s in the early subset to ~27 m/s in 2008–2011 data, and a periodogram of the later subset shows no peak at the orbital frequency. A frequency analysis of the full RV set yields periods of roughly 488 and 530 days, which the authors interpret as beating that modulates the observed amplitude. They also analyze COBE/DIRBE 1.25 micron photometry and report a peak at the planet's orbital period plus a 170-day period. The paper concludes that the planet hypothesis is conclusively ruled out and that the variations are intrinsic stellar oscillations, possibly oscillatory convection modes, with implications for the statistics of planets around K giant stars.","tokens_in":14766,"tokens_out":4776,"duration_ms":50333,"significance":"If the core conclusion holds, the paper removes a widely cited exoplanet from the literature and strengthens the emerging caution that long-period RV signals in K giants can be mimicked by intrinsic stellar variability. The authors explicitly test the detectability of the 479-day signal in the later subset by injecting a synthetic orbit with the same sampling, which is a constructive and falsifiable check. The comparison of RV amplitudes between two well-sampled epochs is a strong, simple argument against a Keplerian companion. However, the paper currently overstates the certainty of the photometric evidence, which is important because the abstract and Section 4.4 assign the word 'conclusively' specifically to the DIRBE photometry. The manuscript is therefore significant but needs revision to bring the claims in line with the actual support.","major_comments":[{"comment":"The DIRBE photometric analysis is presented as conclusively ruling out the planet, but the manuscript never states the time baseline or sampling of the DIRBE 1.25 micron time series. The Price et al. (2010) data are weekly averages from the cryogenic COBE mission, whose useful DIRBE data span only about ten months, shorter than the 479-day period. In a periodogram of a time series shorter than the candidate period, a broad low-frequency peak can arise from a monotonic or low-order trend; the windowed bootstrap as described tests white noise at the chosen frequency but does not test the null of a low-frequency trend. The phased data in Figure 7 cover only a fraction of a cycle, making a sine fit nearly indistinguishable from a linear drift. The authors should report the exact baseline, detrending or trend tests, and then downgrade the photometric claim from 'conclusive' to supporting evidence.","section":"§4.4"},{"comment":"The frequency analysis of the full RV data set does not account for the instrumental zero-point offset that the authors explicitly state exists between the 2004–2011 and 2014–2018 measurements. If the offset is not modeled, a two-segment time series with different means can inject power at low frequencies and can bias the recovered periods and amplitudes of the 487.8-day and 534.8-day signals. The manuscript should either include a fitted zero-point offset in the pre-whitening procedure or demonstrate that the recovered periods are unchanged when the offset is marginalized.","section":"§4.2"},{"comment":"The central amplitude-drop argument needs more detail and internal consistency. The text refers to 'JD = 245660' when the intended boundary is presumably JD 2454660, and the caption of Figure 3 gives a different range. The statement that two data points were removed 'in order to provide a larger time gap between the two subset data' requires specification of which points were removed and a sensitivity test, because the K-amplitude comparison is load-bearing for the refutation. The authors should also show that fitting K alone, with all other orbital parameters fixed, is not biased by phase coverage or by the choice of the adopted orbital solution.","section":"§4.1"}],"minor_comments":[{"comment":"In the bootstrap description, the text says 'the RV data are randomly shuffled' but the analysis is applied to the DIRBE photometry; this should be corrected.","section":"§4.4"},{"comment":"The pre-whitening description should state the frequency grid, the window-function treatment, and the number of significant peaks checked beside the three listed in Table 5; the 534.8-day period is close to the 487.8-day period and a beat interpretation should be checked against aliases.","section":"§4.2"},{"comment":"The abstract states the second period as 530 days while Table 5 lists 534.8 ± 2.2 days; the text should use a consistent value or clearly explain the rounding.","section":"Abstract / §5"},{"comment":"The Hill radius formula is written as rH ≈ a (m / 3M)^(1/3) without parentheses; it should be rH ≈ a (m / (3M))^(1/3).","section":"§5"},{"comment":"There are several typographical issues, including 'firmly established' in the Conclusions, 'these that we are seeing stellar oscillations', and the inconsistent JD boundary in §4.1; a careful language pass is needed.","section":"General"},{"comment":"The caption should explicitly indicate the data subsets used in the upper and lower panels and state how the two removed points were selected.","section":"Figure 3"}],"recommendation":"major_revision","confidential_remarks":"The referee report highlights that the DIRBE claim is not currently supported because the time baseline is never stated and is likely shorter than the 479-day period. The RV evidence for refutation is considerably stronger, so the paper can likely be repaired by tempering the abstract and conclusions, adding the missing baseline information, and addressing the zero-point offset in the full-data frequency analysis. I would not reject the manuscript, but the current wording overreaches and the revision should be substantive rather than cosmetic."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Colleague,\n\nThe bottom line: this paper makes a convincing case that 42 Dra b does not exist, and it does so mainly with radial velocities, not the photometry that the abstract leans on. The RV amplitude of the 479-day signal drops from about 96 m/s to 27 m/s between the 2004-2008 and 2008-2011 subsets. A stable Keplerian orbit cannot produce that. Their simulation shows that even with the sparse sampling of the later subset, the published orbit would have been detected at constant amplitude. That is a clean, quantitative refutation.\n\nThe paper also finds two closely spaced periods in the full 15-year data set (487.8 and 534.8 days) and argues they would be dynamically unstable if both were planetary. That is a reasonable supporting argument, though not as airtight as the amplitude drop.\n\nWhere the paper overshoots is the COBE/DIRBE photometry. Section 4.4 claims the planet hypothesis is 'conclusively ruled out' because the 1.25 micron light curve shows a peak at 479 days with a bootstrap FAP of about 2e-3. But the paper never states the time baseline of the DIRBE data. The COBE cryogenic mission ran roughly ten months, shorter than 479 days. A low-frequency trend over that span can produce periodogram power at a period near the baseline length, and the windowed bootstrap tests white noise, not a trend. So the 'final refutation' is not secure unless they report the baseline and detrend the data. The RV evidence stands on its own, so this is a revision issue, not a rejection.\n\nMinor concern: the zero-point offset between the 2004-2011 and 2014-2018 RV sets is acknowledged in Section 3 but not visibly modeled in the frequency analysis. They should say how they handled it. The oscillatory convection mode speculation is clearly labeled as a hypothesis, so that does not bother me.\n\nWho is this for? Anyone working on RV planet confirmation or occurrence rates around intermediate-mass stars. It is a useful cautionary case and deserves a serious referee. I would accept it for review with revisions.","headline":"The RV amplitude drop is a convincing refutation of 42 Dra b, but the DIRBE 'conclusive' claim overreaches and needs a baseline check.","tokens_in":15345,"tokens_out":3946,"would_cite":true,"duration_ms":36687,"reading_group":"yes","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":[],"pacs":[],"model":"deepseek-v4-flash","headline":"The claimed 3.9-Jupiter-mass planet in a 479-day orbit around 42 Draconis is not there; the radial-velocity signal faded and infrared photometry varies at the same period, so the variations are intrinsic stellar oscillations.","keywords":["42 Draconis","K giant stars","radial velocity planets","exoplanet false positives","stellar oscillations","COBE/DIRBE photometry","long-period variability","planet retraction"],"falsifier":"Check the first and last epochs of 42 Dra in the COBE/DIRBE 1.25-micron photometry: if they span less than 479 days, the peak at the orbital frequency in Figure 6 is not a confirmed periodic signal. Alternatively, obtain fresh high-cadence radial velocities over two or more 479-day cycles; if the signal's amplitude and phase remain coherent, the planet hypothesis would be revived.","tokens_in":14185,"feed_emoji":"⭐","tokens_out":10297,"duration_ms":94973,"temperature":0.7,"pith_summary":"This paper sets out to show that the giant planet previously reported around the K giant star 42 Draconis does not exist. Fifteen years of radial-velocity measurements show the 479-day signal attributed to a 3.9-Jupiter-mass companion did not stay coherent: its amplitude fell by a factor of about four, and a second period near 530 days appears in the later data, so the two close periods can beat and mimic a decaying signal. Independent near-infrared photometry from the COBE/DIRBE archive varies at the proposed orbital period and at a second period of about 170 days, which the authors take as direct evidence that the variations are intrinsic to the star rather than orbital reflex motion. If correct, the result adds 42 Dra to a list of long-period K-giant 'planet' signals that vanish under longer monitoring, with consequences for planet-occurrence statistics around intermediate-mass stars.","feed_headline":"No giant planet orbits 42 Draconis; the RV signal was the star","feed_subtitle":"Fifteen years of radial velocities plus infrared photometry show the 479-day signal was an oscillation, not a companion.","key_machinery":"The central mechanism is beating between two closely spaced radial-velocity periods, 487.8 and 534.8 days, whose superposition makes a single roughly 479-day signal appear to shrink and fade over time. The analysis runs on Lomb-Scargle periodograms with pre-whitening, on the bisector velocity span of the cross-correlation function, and on the COBE/DIRBE 1.25-micron photometric time series, which supplies the independent evidence that the star itself varies at the proposed orbital period and at 169.8 days. The Hill stability criterion seals the interpretation: two hypothetical planets at 1.24 and 1.32 AU would be too close for a stable system, so the two periods cannot be a compact planetary pair.","core_discovery":"The paper's central claim is that 42 Draconis hosts no giant planet. The original detection, a 479.1-day radial-velocity signal with a semi-amplitude near 110 m/s and a minimum mass of 3.9 Jupiter masses, survived the standard checks: no H-alpha correlation, no line-shape variations at the orbital period, and no Hipparcos photometric signal at that period. Continued monitoring broke it: the 2008-2011 data alone show no peak at 479 days, and a fit that fixes all orbital parameters except amplitude gives K = 27.4 m/s compared with 96.6 m/s for the earlier subset. A frequency analysis of the full 2004-2018 dataset finds periods of 487.8 and 534.8 days whose beating can reproduce the amplitude decline, and 1.25-micron COBE/DIRBE photometry shows significant power at the orbital period and at 169.8 days. These two periods cannot be two planets: at their masses and separations the system would be dynamically unstable by the Hill criterion. The authors conclude that the variations are stellar oscillations, most likely oscillatory convection modes.","pith_inferences":["The paper calls the DIRBE evidence conclusive but never states the time baseline of the photometry; if that baseline is shorter than 479 days, the peak at the orbital frequency could be a low-frequency trend, so the independent photometric refutation would need re-examination.","The same DIRBE-based search could be applied to other K giants with claimed long-period planets; a star whose infrared photometry varies at the RV period would be a strong candidate for retraction.","The 530-day and 294-day periods found here could be a signature of the suspected oscillation modes, testable by looking for similar period ratios in other luminous K giants.","If Gaia astrometry resolves the motion of 42 Dra over the same epochs, a real companion would leave a measurable acceleration that the oscillation interpretation would not, giving a dynamical test independent of photometry."],"forward_implications":["The named exoplanet 42 Dra b (Orbitar) would have to be retracted from planet lists.","Long-period radial-velocity signals around K giants should not be regarded as confirmed planets unless monitoring spans more than a decade and checks for amplitude and phase changes.","Photometric variation at the proposed orbital period is a decisive test: when the star itself brightens and dims on the 'planet' period, the companion interpretation is ruled out.","Planet-occurrence statistics for intermediate-mass stars may be inflated by this class of false positives, potentially biasing formation theories.","If oscillatory convection modes are confirmed, they become a known source of long-period RV variability that future surveys will need to account for."],"supporting_citations":[{"why":"Original discovery of the 479-day, 3.9 Jupiter-mass planet signal around 42 Dra; the claim this paper refutes.","marker":"D09"},{"why":"Source of the COBE/DIRBE 1.25-micron photometry time series used to show variability at the planet period and at 169.8 days.","marker":"Price et al. (2010)"},{"why":"Lomb-Scargle periodogram method used for all frequency analyses in the paper.","marker":"Scargle (1982)"},{"why":"Scaling relations used to show the short-term 1.3-day and 0.82-day RV variations match p-mode oscillation amplitudes and frequencies.","marker":"Kjeldsen & Bedding (1995)"},{"why":"Hill stability criterion used to show the hypothetical two-planet system at 1.24 and 1.32 AU would be dynamically unstable.","marker":"Gladman (1993)"},{"why":"Provides the gamma Dra precedent of a long-period RV 'planet' that vanished, supporting the oscillatory-convection interpretation.","marker":"Hatzes et al. (2018)"},{"why":"Called the Aldebaran planet into question through additional RV measurements, the pattern this paper applies to 42 Dra.","marker":"Reichert et al. (2019)"}],"fun_headline_variants":["42 Draconis exoplanet ruled out: signal is stellar oscillation","No planet around 42 Dra: RV signal was stellar oscillation","42 Dra's planet signal explained as stellar oscillation","No giant planet for 42 Dra; RV variation is stellar oscillation","42 Dra ruled out as host of giant planet; signal is oscillation"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The refutation's strongest independent evidence assumes the COBE/DIRBE 1.25-micron time series is long enough and well enough sampled to constrain a 479-day period, but the paper never states that baseline, and a shorter baseline could make the orbital-frequency peak a low-frequency trend.","fun_headline_variants_meta":{"raw":{"variants":["42 Draconis exoplanet ruled out: signal is stellar oscillation","No planet around 42 Dra: RV signal was stellar oscillation","42 Dra's planet signal explained as stellar oscillation","No giant planet for 42 Dra; RV variation is stellar oscillation","42 Dra ruled out as host of giant planet; signal is oscillation"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000977,"raw_usage":{"total_tokens":4242,"prompt_tokens":1126,"completion_tokens":3116,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":742,"completion_tokens_details":{"reasoning_tokens":3029}},"tokens_in":742,"tokens_out":3116,"duration_ms":21310,"temperature":1.0,"reasoning_tokens":3029,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-15T23:09:27.327827+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Check the first and last epochs of 42 Dra in the COBE/DIRBE 1.25-micron photometry: if they span less than 479 days, the peak at the orbital frequency in Figure 6 is not a confirmed periodic signal. Alternatively, obtain fresh high-cadence radial velocities over two or more 479-day cycles; if the signal's amplitude and phase remain coherent, the planet hypothesis would be revived.","supporting_citations":[{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Source of the COBE/DIRBE 1.25-micron photometry time series used to show variability at the planet period and at 169.8 days."},{"cited_title":"& Bedding, T.R","cited_arxiv_id":null,"evidence_quote":"Scaling relations used to show the short-term 1.3-day and 0.82-day RV variations match p-mode oscillation amplitudes and frequencies."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Provides the gamma Dra precedent of a long-period RV 'planet' that vanished, supporting the oscillatory-convection interpretation."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Called the Aldebaran planet into question through additional RV measurements, the pattern this paper applies to 42 Dra."}],"review_version":1}