{"id":"3ef28314-5232-4788-b5c8-85b846552541","arxiv_id":"2411.14911","paper_version":2,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":6.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":5,"one_line_summary":"First RV masses for TOI-396 b and d, significant TTVs for b and c, and a demonstration that the inner pair is near but not in the 5:3 resonance.","lead":"Astronomers measured the masses of two planets in the bright TOI-396 system and found the outermost planet is the densest. They also detected transit timing variations that show the inner pair is near, but not in, a 5:3 orbital resonance.","discovery_kind":"extension","skeptic_critique":{"model":"deepseek-v4-flash","headline":"The e=0 assumption is backed by a ΔBIC>100 from eccentric runs that the authors report as poorly converged; with TRADES finding e≈0.08, the circular-fit masses of b and d could shift.","rationale":"The paper is careful and well-structured; the central RV detection of b and d is plausible and supported by consistency with the TRADES dynamical masses. The TTV significance is supported by reduced chi-square values and the anti-correlation pattern, and the authors appropriately hedge the TTV mass of c. The weakest point in the chain is the dismissal of eccentric orbits: the reported ΔBIC>100 is difficult to reconcile with the authors' own statement that the eccentric MCMC runs converged poorly, and the same eccentricity range (e≈0.08) is later preferred by the dynamical fit. Since the headline masses are derived from a circular model with K values only slightly above the RV jitter, a quantitative check with an informative eccentricity prior is the single most decisive test. The reader's conditional verdict remains appropriate; this check would either confirm the robustness of the masses or require that the mass claims be softened.","tokens_in":39517,"tokens_out":13383,"duration_ms":140347,"concrete_test":"Re-run the Section 5.2 joint LC+RV MCMC with eccentricities free under the same half-Gaussian prior used in the TRADES fit (e=0, σ=0.083) instead of wide uniform bounds; compare K_b, K_d, and ΔBIC with Table 2. If K_b or K_d moves by more than ~1σ, or if the eccentric model is not strongly disfavoured, the circular-derived masses are model-dependent and the headline claims should be presented as conditional on the circular-orbit assumption.","verdict_should_be":"UNCHANGED","load_bearing_attack":"Section 4 rejects eccentric orbits primarily via ΔBIC = BIC(e≠0) − BIC(e=0) ≳ +100, but the same paragraph states that the eccentric runs have 'the poorer the parameter convergence' and that the data 'are not enough to constrain the planetary eccentricities well.' A converged eccentric fit can only add the BIC penalty for its extra parameters (roughly +30 for six parameters, not +100); a +100 deficit therefore signals an unreliable or incomplete comparison, not a clean circular-orbit preference. In contrast, the TRADES dynamical analysis in Section 6, using a half-Gaussian prior with σ=0.083, returns e≈0.08 for all three planets. Because K_b≈1.30 and K_d≈1.78 m/s are comparable to the 1.49 m/s RV jitter, even a modest eccentricity-induced bias in K can move the headline masses Mb=3.55 and Md=7.1 M⊕ by more than their formal uncertainties. The paper's injection-recovery tests target P_c and P_rot, but not the orbital periods of b and d, so the stability of K_b and K_d under realistic eccentricities is not demonstrated.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper presents a joint analysis of TESS transit photometry and HARPS radial velocities of the bright F6V star TOI-396, which hosts three transiting planets near 2 Earth radii. Using MCMC joint fits with breakpoint-based activity detrending, the authors refine the planet radii (Rb=2.004+0.045-0.047, Rc=1.979+0.054-0.051, Rd=2.001+0.063-0.064 Rearth), report first RV mass detections for b (Mb=3.55+0.94-0.96 Mearth) and d (Md=7.1+/-1.6 Mearth), and a 3-sigma upper limit for c (Mup,c=3.8 Mearth). They identify significant TTVs for b and c with an anti-correlated pattern, perform a TRADES dynamical analysis yielding a formally precise but explicitly caveated mass for c (Mc,dyn=2.24+0.13-0.67 Mearth), find the b-c pair is near but not inside the 5:3 MMR, and forecast TTV amplitudes up to ~5 hours over a 5.2-year baseline. They also simulate JWST eclipse observations. The paper is transparent about the limitations of the TTV-derived mass and about the RV non-detection of c, attributing the latter to proximity of Pc to the stellar rotation period (Prot=6.7+/-1.3 d).","tokens_in":39774,"tokens_out":6376,"duration_ms":63646,"significance":"If the measured masses hold, the system is unusual in that the outermost planet is the densest, and the very bright host star (V~6.4) makes TOI-396 an important benchmark for formation models and JWST atmospheric characterization. The paper is methodologically careful in several respects: BIC-based model selection, breakpoint activity correction, explicit jitter modeling, MCMC convergence checks, injection-recovery tests for activity periods, and frank caveats about the TTV mass. The improved radii (~1.4x precision over Vanderburg et al. 2019) and the first RV masses for b and d are concrete contributions. However, the headline masses rest on a circular-orbit assumption whose evidential basis is weaker than the reported ΔBIC implies, and the stability of the small RV semi-amplitudes against the adopted eccentricity and activity model is not demonstrated. These points need to be addressed before the central mass claims can be considered robust.","major_comments":[{"comment":"The circular-orbit assumption is load-bearing for the headline masses (K_b=1.30+0.34-0.35 m/s and K_d=1.78+/-0.40 m/s are comparable to the 1.49 m/s jitter), but the evidence for e=0 is internally inconsistent. The paragraph that rejects eccentric orbits states both that the eccentric MCMC runs have \"the poorer the parameter convergence\" and that the data \"are not enough to constrain the planetary eccentricities well,\" yet it then uses ΔBIC = BIC(e!=0)-BIC(e=0) ≳ +100 to discard them. For six additional parameters, a converged eccentric fit would carry a BIC penalty of only ~ln(N) per parameter (~26 for 78 RVs); a ΔBIC of +100 therefore implies either a large chi-square difference that is hard to believe given the stated convergence problems, or an unreliable comparison. The TRADES dynamical fit in Section 6 returns e≈0.08 for all three planets (with a half-Gaussian prior, sigma=0.083), which is in tension with a strongly circular solution. Because the RV semi-amplitudes are small, a bias of even a few tenths of m/s from unmodeled eccentricity could shift Mb and Md by more than their formal errors. Please provide a more robust test than the reported ΔBIC, for example a fixed-eccentricity grid for e in [0, 0.15], an injection-recovery study at K_b and K_d with e≈0.08, or a rerun of the eccentric fit with a better-converging sampler/parameterization, and show quantitatively how K_b and K_d, hence Mb and Md, depend on the eccentricity assumption.","section":"Section 4 and Table 2"},{"comment":"The injection-recovery tests are used to argue that stellar activity explains the RV non-detection of planet c and to quantify signal suppression at P_rot. However, they do not test the recovery of the small semi-amplitudes of b and d under the adopted activity model and breakpoint detrending. Since K_b and K_d are only 3.8σ and 4.5σ detections and are of the same order as the RV jitter, an injection test at Pb and Pd (with and without the breakpoint model) would show whether the fitted K values are biased by the de-trending procedure. In addition, the injection tests assume circular orbits; given major comment 1, repeating them for e≈0.08 would directly quantify the potential mass bias. Without such tests, the quoted mass uncertainties (Table 2) likely understate the true systematic error budget.","section":"Section 5.3 and Tables 3-4"}],"minor_comments":[{"comment":"The text in Section 5.1 gives 3σ upper limits K_c^up = 1.2 m/s, M_c^up = 4.0 Mearth, and rho_c^up = 3.1 g/cm3, while Table 2 note (c) gives M_c^up = 3.8 Mearth and rho_c^up = 2.9 g/cm3; please harmonize these values.","section":"Section 5.1 vs Table 2"},{"comment":"The sentence \"imposing uniform priors on (sqrt(e) cos(omega), sqrt(e) cos(omega))\" contains a typo; the second coordinate should be sqrt(e) sin(omega).","section":"Section 4"},{"comment":"The statement that the recovery of injected signals at P_c implies that \"the destructive interference between the RV signals induced by the star and by planet c has already occurred\" is not the only possible interpretation; the original K_c may simply be small because planet c is low-mass. Please clarify the logic or soften the wording, especially since the upper limit Mup,c=3.8 Mearth allows a low-mass c.","section":"Section 5.3"},{"comment":"The TTV forecast with semi-amplitudes up to ~5 hours (and drifts up to ~10 hours) is derived from the MAP parameters fitted to the same TTV data; the paper correctly presents this as a model forecast, but the abstract and conclusions could more explicitly state that the amplitude prediction is conditional on the dynamical model and the poorly sampled TTV phase, not an observationally measured amplitude.","section":"Section 6 and Figure 9"}],"recommendation":"major_revision","confidential_remarks":"The manuscript is a solid and unusually honest characterization paper. The main weakness is that the headline RV masses are anchored to a circular-orbit assumption whose statistical support is undermined by the authors' own statement that the eccentric runs do not converge; the reported ΔBIC>+100 is thus not a reliable discriminator. I would like to see either a better-converged eccentric fit, a fixed-e grid, or an injection-recovery test at K_b and K_d with e≈0.08 before the mass claims can be considered robust. The paper's transparency about the TTV mass and the non-detection of c is commendable and should be preserved."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"This is a solid, useful paper. The new results are real: first RV masses for TOI-396 b and d, radii improved by ~1.4x with two extra TESS sectors, and a significant TTV detection for b and c. The non-resonance conclusion now rests on dynamical angle evolution, not just period ratios, and the anti-correlated TTV pattern is a nice confirmation that c is a real planet despite being invisible in the RVs.\n\nThe analysis is careful. They use BIC-based model selection, breakpoint activity correction, jitter modeling, injection tests, and convergence checks. They also handle the awkward TTV mass of c the right way: they report Mc,dyn = 2.24 M⊕ but explicitly warn that it might not be accurate because the TTV phase is poorly sampled. That kind of honesty is worth crediting.\n\nThe main soft spot is the fixed-eccentricity assumption in the mass fit. The paper rejects eccentric orbits mainly via a reported ΔBIC > 100, but admits those eccentric runs had poor parameter convergence. That is not a clean rejection. And their own TRADES dynamical analysis, even with a half-Gaussian prior pulling toward zero, finds e ≈ 0.08 for all three planets. With K_b and K_d at 1.3 and 1.8 m/s against 1.49 m/s jitter, a modest eccentricity bias could shift the headline masses by more than their formal errors. The injection tests target the rotation period and P_c, not the b and d orbital periods, so the stability of those semi-amplitudes is not demonstrated. I would want the eccentricity issue addressed properly, either with a converged eccentric fit or with injections at Pb and Pd. It may not change the qualitative picture, but it matters for the claimed precision.\n\nA minor point: no analysis scripts are provided, so the detailed steps are hard to reproduce from the paper alone. The data tables in the appendix help, but scripts would be better.\n\nOverall, the central claims are supported and the weaknesses are addressable rather than fatal. This is a benchmark system worth having in the literature, and it deserves a serious referee. I would recommend engaging with it, and I would be happy to cite the masses and radii once the eccentricity question is cleaned up.","headline":"Solid benchmark characterization of a bright multi-planet system; the new RV masses for b and d are probably right, with caveats on the eccentricity assumption and the TTV-derived mass of c.","tokens_in":40556,"tokens_out":1602,"would_cite":true,"duration_ms":19144,"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 three-planet system TOI-396 has nearly equal planet radii, but the newly measured masses show the outermost planet is the densest, and the inner pair's transit-timing variations place it close to, but outside, the 5:3 mean-motion…","keywords":["TOI-396","transit-timing variations","mean motion resonance","radial velocities","stellar activity","planetary masses","TESS photometry","HARPS spectroscopy"],"falsifier":"Full TTV phase coverage over the predicted ~5-year super-period would settle the dynamical picture: if the transit times do not show the modeled drift of up to about 5 hours for b and 10 hours for c, or if a Keplerian signal near 5.97 days emerges in a longer activity-calibrated RV baseline, the paper's TTV solution and its explanation for the RV non-detection of planet c would be ruled out.","tokens_in":39324,"feed_emoji":"🪐","tokens_out":11340,"duration_ms":94734,"temperature":0.7,"pith_summary":"This paper establishes the architecture of TOI-396, the brightest known star hosting three transiting planets. Combining four TESS sectors with HARPS radial velocities, the authors refine the three radii to near-identical values around 2 Earth radii and, for the first time, measure Doppler masses for the inner and outer planets: TOI-396 b at 3.55 Earth masses and TOI-396 d at 7.1 Earth masses, making the outermost planet the densest. The middle planet, TOI-396 c, is invisible in the radial-velocity data because its 5.97-day orbit nearly coincides with the star's 6.7-day rotation period, but significant and anti-correlated transit-timing variations confirm its gravitational presence. A dynamical fit and a 10,000-year integration of the resonance angles show that planets b and c are close to, but not inside, the 5:3 mean-motion resonance. If these results hold, the system becomes a bright, nearby laboratory for testing how equal-size planets acquire very different densities.","feed_headline":"Masses measured for two planets; outer is the densest","feed_subtitle":"New masses for TOI-396 b and d; transit-timing wobbles put the inner pair just outside the 5:3 resonance.","key_machinery":"The argument is carried by a joint analysis of 41 TESS transit light curves and HARPS radial velocities in an MCMC framework, with stellar activity removed from the RVs by a breakpoint method that splits the time series at a statistically selected epoch and de-trends each segment separately. The resonance question is decided by the critical angles $\\phi_b = 3\\lambda_b - 5\\lambda_c + 2\\varpi_b$ and $\\phi_c = 3\\lambda_b - 5\\lambda_c + 2\\varpi_c$ of the 5:3 mean-motion resonance: if these angles librate the pair is trapped in resonance, and if they circulate it is not. An N-body dynamical fit to the RVs and transit times supplies the planet masses from the TTVs, while forward integrations predict how the transit-timing amplitudes grow with time.","core_discovery":"On the paper's own terms, the central discovery is that the three TOI-396 planets have essentially equal radii—$R_b = 2.004^{+0.045}_{-0.047}\\,R_\\oplus$, $R_c = 1.979^{+0.054}_{-0.051}\\,R_\\oplus$, and $R_d = 2.001^{+0.063}_{-0.064}\\,R_\\oplus$—while their masses differ: the first RV-based mass determinations give $M_b = 3.55^{+0.94}_{-0.96}\\,M_\\oplus$ and $M_d = 7.1 \\pm 1.6\\,M_\\oplus$, with bulk densities of $2.44$ and $4.9$ g cm$^{-3}$. The outermost planet being the densest is an unusual architecture. TOI-396 c remains undetected in the RVs, and the paper argues this is because its period is too close to the stellar rotation period; instead, its presence is established by significant, anticorrelated transit-timing variations in planets b and c. The dynamical analysis yields a formally precise mass for c of $M_{c,\\mathrm{dyn}} = 2.24^{+0.13}_{-0.67}\\,M_\\oplus$, which the authors caution may be inaccurate until the TTV phase is fully sampled, and it shows the b–c pair is close to but out of the 5:3 mean-motion resonance.","pith_inferences":["If the dynamical mass of planet c near 2.2 Earth masses survives full TTV sampling, the three planets would cover a factor-of-three mass range at nearly identical radii, making TOI-396 a sharper test of radius plateau models than a single-mass target.","The same activity-suppression effect demonstrated here may apply to other RV surveys: planets with periods within about one day of the stellar rotation period could be systematically missing from Doppler mass catalogs, biasing demographic conclusions.","A dedicated campaign spanning the predicted ~5-year TTV super-period could turn TOI-396 b and c into a precise dynamical clock, potentially revealing additional companions or tidal effects through deviations from the current N-body model.","The anti-correlated TTVs already provide an independent, photometric confirmation that planet c exists; this supports the use of TTVs as a discovery channel for planets hidden from RV surveys by activity."],"forward_implications":["If the architecture is real, TOI-396 joins the small set of systems where three similar-radius planets have resolved density differences, directly testing 'peas in a pod' formation models.","Future transit observations of planets b and c must account for predicted timing drifts of up to roughly 5 and 10 hours, respectively, relative to the linear ephemeris.","Completing the TTV phase coverage—for example with additional TESS or CHEOPS photometry—should turn the formal dynamical mass of planet c into a reliable one.","The injection tests imply that any planet with an orbital period near the stellar rotation period will have its RV signal systematically suppressed, so non-detections in such cases are not strong mass upper limits.","JWST eclipse observations with 2, 4, and 8 events for planets b, c, and d should distinguish primary from secondary atmospheres at the 3σ level, linking bulk density to atmospheric composition."],"supporting_citations":[{"why":"Discovered the three transiting planets, supplied the first radii and periods, excluded false positives, and simulated low-eccentricity configurations that this paper refines.","marker":"Vanderburg et al. (2019)"},{"why":"Provides the skew-normal fit to the HARPS cross-correlation functions used to extract the radial velocities and activity indicators.","marker":"Simola et al. (2019)"},{"why":"Provides the breakpoint method that splits the RV time series into segments so stellar activity can be de-trended piecewise.","marker":"Simola et al. (2022)"},{"why":"Supplies the MCMC code used for the joint analysis of TESS transit light curves and HARPS radial velocities.","marker":"Bonfanti & Gillon (2020)"},{"why":"Supplies the N-body dynamical framework used for the joint TTV and RV fit that yields the dynamical mass of planet c.","marker":"Borsato et al. (2014, 2019, 2021)"},{"why":"Provides the long-term integrator used to evolve the system for 10,000 years and test libration versus circulation of the resonance angles.","marker":"Rein & Liu (2012); Rein & Tamayo (2015)"},{"why":"Motivates the rebinning of the 30-minute-cadence Sector 3 light curves so the transit parameters are not biased by long cadence.","marker":"Kipping (2010)"}],"fun_headline_variants":["Equal radii, unequal masses: outer TOI-396 planet is densest","TTVs show inner pair near 5:3 resonance, not in it","First RV masses for TOI-396 b and d: outer is denser","Three equal-size planets, but the farthest is heaviest"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The load-bearing assumption is that all three planets follow circular orbits; if the near-resonant pair b and c has non-negligible eccentricity, the fitted radial-velocity amplitudes and thus the derived masses could be biased.","fun_headline_variants_meta":{"raw":{"variants":["Equal radii, unequal masses: outer TOI-396 planet is densest","TTVs show inner pair near 5:3 resonance, not in it","First RV masses for TOI-396 b and d: outer is denser","Three equal-size planets, but the farthest is heaviest"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.00073,"raw_usage":{"total_tokens":3491,"prompt_tokens":1391,"completion_tokens":2100,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":1007,"completion_tokens_details":{"reasoning_tokens":2020}},"tokens_in":1007,"tokens_out":2100,"duration_ms":14511,"temperature":1.0,"reasoning_tokens":2020,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-12T14:43:13.129048+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Full TTV phase coverage over the predicted ~5-year super-period would settle the dynamical picture: if the transit times do not show the modeled drift of up to about 5 hours for b and 10 hours for c, or if a Keplerian signal near 5.97 days emerges in a longer activity-calibrated RV baseline, the paper's TTV solution and its explanation for the RV non-detection of planet c would be ruled out.","supporting_citations":[{"cited_title":"X., Rodriguez, J","cited_arxiv_id":null,"evidence_quote":"Discovered the three transiting planets, supplied the first radii and periods, excluded false positives, and simulated low-eccentricity configurations that this paper refines."},{"cited_title":"2019, A&A, 622, A131","cited_arxiv_id":null,"evidence_quote":"Provides the skew-normal fit to the HARPS cross-correlation functions used to extract the radial velocities and activity indicators."},{"cited_title":"2022, A&A, 664, A127","cited_arxiv_id":null,"evidence_quote":"Provides the breakpoint method that splits the RV time series into segments so stellar activity can be de-trended piecewise."}],"review_version":1}