{"id":"fd6023b6-7115-40f3-b006-102744edca8d","arxiv_id":"2504.16159","paper_version":1,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":6.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":3,"one_line_summary":"The Perseus star-forming complex is probably not dispersing; its apparent expansion is a projection effect of clusters orbiting the Galaxy, and LISCA I and W345 formed a few hundred parsecs apart about 20 to 30 Myr ago.","lead":"Using Gaia measurements of star clusters in the W3/W4/W5 star-forming region and the wider Perseus complex, the authors reconstruct the region's three-dimensional motions and orbital histories. They argue that the apparent outward expansion seen in earlier data is mostly a projection effect, with clusters orbiting the Galaxy at slightly different speeds rather than flying apart.","discovery_kind":"new_application","skeptic_critique":{"model":"deepseek-v4-flash","headline":"The no-dispersal and LISCA I-W345 co-formation conclusions depend on a line-of-sight velocity choice for SAI 24 where two catalogs disagree at 3.7 sigma and the paper adopts the value consistent with the W345 mean.","rationale":"The reader's weakest assumption identifies the same load-bearing spot: the adopted SAI 24 line-of-sight velocity. This is the right concern because the paper's headline claims about no dispersal and shared formation history are not derived from the internal kinematics of individual clusters but from backward orbit integrations of aggregate mean phase-space coordinates, and vLOS is the least secure component of those initial conditions. The paper is honest about the discrepancy, but the stated reason for choosing Hunt and Reffert (2023) over Tarricq et al. (2021) is consistency with the W345 mean velocity, which introduces a mild circularity into the aggregate average and into the orbit that is then used to infer the complex's history. A 25 km/s shift in the W345 mean vLOS is large compared with the spread among W345 clusters and can plausibly move the LISCA I-W345 encounter epoch beyond the quoted 20-30 Myr window. The direct 3D regression in Fig. 10 also gives 7 +/- 10 km/s/kpc, which is consistent with zero but does not by itself exclude mild expansion; the stronger 'not dispersing' phrasing in the abstract therefore leans on the orbit integrations, making the SAI 24 choice load-bearing. The spiral-arm amplitude f is unconstrained, but the authors explore f = 0.1, 0.3, 0.5 and explicitly caveat it, so it is not the weakest point. Independent support is limited but real: the cluster samples are built from Gaia DR3 with homogeneous astrometric selections, the orbit integrations use public AGAMA machinery with error sampling, and the authors verify that results do not change when using the Hunt and Reffert catalog for distances. None of this removes the need for the concrete alternative-vLOS rerun. Because the reader already assigned CONDITIONAL and the concern is exactly the one motivating that verdict, no verdict adjustment is needed; the stress-test confirms the conditionality rather than overturning it.","tokens_in":31824,"tokens_out":3793,"duration_ms":38706,"concrete_test":"Re-run the Section 5.3 aggregate construction and backward orbit integrations with vLOS(SAI 24) = +52 +/- 22 km/s from Tarricq et al. (2021) instead of the adopted Hunt and Reffert value, keeping all other cluster data and sampling identical; also run a variant that excludes SAI 24 from the W345 aggregate mean. Compare the W345-LISCA I inter-aggregate distance curve in Fig. 14: if the minimum distance at t about -20 to -30 Myr remains a few hundred parsecs within the 16th-84th percentile bands, the co-formation claim survives this test; if the minimum shifts by more than the quoted uncertainties or disappears, the central no-dispersal claim depends on the disputed SAI 24 vLOS and should be presented as conditional.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The central claim that the Perseus complex is not dispersing and that LISCA I and W345 formed 20-30 Myr ago only a few hundred parsecs apart is established through backward orbit integrations of aggregate phase-space coordinates (Section 5.3, Figs. 12-14). Those integrations are most sensitive to the adopted line-of-sight velocities, and the most fragile entry is SAI 24. In Section 5.1, Tarricq et al. (2021) report vLOS = +52 +/- 22 km/s from 8 stars, while Hunt and Reffert (2023) report vLOS = -48 +/- 5 km/s from 7 stars. The authors acknowledge the 3.7-sigma discrepancy and state they adopt the Hunt and Reffert value because it is closer to the W345 mean velocity around -40 km/s. This is a selection based on consistency with the quantity being averaged, not on an independent validation of one measurement. SAI 24 enters the W345 aggregate: with the adopted value the four-cluster W345 mean vLOS is about -47 km/s, while replacing it with the Tarricq value shifts the mean to about -22 km/s. Since vLOS is explicitly identified as a primary uncertainty source in the orbit integrations, the claimed LISCA I-W345 approach epoch and the conclusion that the observed expansion is a projection effect are contingent on this disputed choice. The missing vLOS measurements for UBC 420, Basel 10, and NGC 637 are secondary but compound the sensitivity of the aggregate means.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper presents a multi-scale study of the W3/W4/W5 (W345) region and the larger Perseus complex using Gaia DR3 and complementary catalogs. The authors identify five known clusters in W345, characterize their structure, ages, and internal kinematics, and find that the three youngest clusters show expansion while the two older ones are near equilibrium. They then assemble 6D phase-space data for 15 clusters across W345, LISCA I, and LISCA II, and integrate cluster aggregate orbits in the axisymmetric McMillan (2017) potential, as well as in models with spiral-arm perturbations and a bar. The central claims are that the Hubble-like expansion reported by Román-Zúñiga et al. (2019) is likely a projection effect caused by different orbital velocities of stars at slightly different Galactocentric distances; that LISCA I and W345 formed 20--30 Myr ago only a few hundred parsecs apart; and that spiral-arm perturbations can keep the aggregates closer for longer.","tokens_in":32098,"tokens_out":3772,"duration_ms":36143,"significance":"If the conclusions hold, the paper provides a novel and important reinterpretation of the kinematics of the Perseus complex, replacing an internal-expansion interpretation with a Galactic-orbit interpretation. The work is careful in several respects: it uses homogeneous astrometric selections and membership criteria, it propagates distance and line-of-sight velocity errors into orbit integrations via Monte Carlo sampling, it makes the AGAMA orbit integrations reproducible, and it explores perturbations of the Galactic potential (spiral arms, bar, varying perturbation strength) in appendices. The cluster characterization in Section 3 is a useful contribution in its own right, and the YSO kinematic analysis in Section 4 adds value. The main weakness is that the dynamical conclusions, especially the no-expansion statement and the LISCA I--W345 co-formation epoch, rest on a small number of line-of-sight velocity measurements, one of which (SAI 24) shows a 3.7-sigma discrepancy between two catalogs. The authors acknowledge this discrepancy but do not demonstrate that their conclusions are robust to the alternative value.","major_comments":[{"comment":"The choice of vLOS for SAI 24 is load-bearing for the paper's central conclusion. Tarricq et al. (2021) report +52 ± 22 km/s while Hunt & Reffert (2023) report -48 ± 5 km/s, and the authors adopt the latter because it is closer to the W345 mean velocity. This selection is based on consistency with the quantity being averaged rather than on an independent validation of either measurement. SAI 24 enters the W345 aggregate; switching to the Tarricq value changes the four-cluster W345 mean vLOS from approximately -47 km/s to approximately -22 km/s. Since the backward orbit integrations in Section 5.3 (Fig. 14) drive the claim of a LISCA I--W345 minimum distance 20--30 Myr ago, the authors should repeat the aggregate mean and orbit integrations using the Tarricq value, and also with SAI 24 excluded, and show whether the conclusions survive.","section":"Section 5.1, Table 4"},{"comment":"The direct 3D distance--velocity regression gives 7 ± 10 km/s/kpc, which is formally consistent with both zero expansion and a mild expansion at the level claimed by Román-Zúñiga et al. (2019). The statement that 'we do not observe a Hubble-like expansion' is therefore not established by this regression alone; it rests on the orbit-integration analysis. Moreover, the regression is performed on absolute 3D velocities in the Galactocentric frame, which include the large common orbital velocity of the complex. The slope could be dominated by the gradient of Galactic rotation across the ~1.2 kpc line-of-sight depth rather than by any internal expansion. I ask the authors to add a test using velocities relative to the complex mean motion, or an explicit forward model of the projection signature, to demonstrate that the absence of a 3D slope is specifically informative about internal dispersal.","section":"Section 5.2, Fig. 10"},{"comment":"Three clusters in Table 4 (UBC 420, Basel 10, and NGC 637) have no line-of-sight velocity measurement in any catalog, yet they are included in the aggregate positions and velocities used for the orbit integrations. The paper does not state which clusters actually contribute to each aggregate's vLOS, nor does it quantify the sensitivity of the aggregate means or the resulting inter-aggregate distances to excluding these clusters (or to assigning them the aggregate mean as a proxy). Given that vLOS is explicitly identified as a primary uncertainty source in the orbit reconstruction, this missing sensitivity analysis compounds the SAI 24 concern and should be reported.","section":"Section 5.1 and 5.3"}],"minor_comments":[{"comment":"There are several typographical and formatting issues: 'Identifing' in the Section 2 title, 'genearlly' in Section 5.3, 'Persues' in Section 5.3, 'Publicy' in a footnote, and frequent 'di fferent' (spurious spaces) throughout the text. These should be corrected in a final pass.","section":"Throughout"},{"comment":"The text says that for clusters not in Tarricq et al. (2021), the authors adopt a mean LOS velocity of -39 km/s from Fanelli et al. (2022b). It would help to state explicitly that this adopted value is used only for the internal expansion analysis (Fig. 8) and not for the orbit integrations in Section 5.3, since the latter require per-cluster vLOS values.","section":"Section 3.3"},{"comment":"The captions and text refer to 'median orbits' but it is not always clear whether the medians are taken across the 500 initial-condition realizations at each time step or across time. A one-sentence clarification in the caption of Fig. 12 would resolve this.","section":"Section 5.3, Fig. 12-14"},{"comment":"The discussion of spiral-arm perturbation strength (f) is clear, but the choice f = 0.3 as the 'reference model' is presented without a quantitative justification beyond the statement that f_HI = 0.53-0.66 is an upper limit. The authors should briefly restate why 0.3 is a reasonable intermediate value for the stellar arm-to-interarm contrast.","section":"Section 5.4"}],"recommendation":"major_revision","confidential_remarks":"The paper is well within the scope of A&A and represents a substantial observational effort. The main risk is the fragility of the dynamical conclusions to one disputed line-of-sight velocity (SAI 24) and to the treatment of clusters lacking vLOS. This is fixable with additional robustness tests and is not grounds for rejection. I do not see any citation or novelty concerns: the prior work by Román-Zúñiga et al. (2019), Dalessandro et al. (2021), and Della Croce et al. (2023) is properly acknowledged and built upon."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Dear colleague,\n\nThis paper is worth taking seriously. It is the first to put the Perseus complex's clusters in 6D and show that the Hubble-like expansion reported by Román-Zúñiga et al. (2019) largely disappears once you account for the complex's ~1.2 kpc line-of-sight depth and differential Galactic rotation. The clusters themselves are all previously known, but the orbit-based interpretation is genuinely new, and the backward integration gives a concrete, quantitative formation scenario: LISCA I and W345 a few hundred parsecs apart 20–30 Myr ago, with LISCA II forming 0.6–1 kpc away.\n\nThe cluster characterization is careful and thorough: density profiles with Plummer fits, structural parameters, ages, internal kinematics, YSO tracing, and ionizing sources. The vLOS compilation is transparent, and the error sampling in the orbit integrations is a real plus. The spiral-arm toy model is explicitly caveated; they test a range of perturbation amplitudes and show that the short-timescale conclusions are robust.\n\nThe soft spot is the one the stress-test flags: SAI 24's vLOS. Tarricq et al. quote +52 ± 22 km/s, Hunt & Reffert quote -48 ± 5 km/s, and the paper adopts the Hunt & Reffert value because it is closer to the W345 mean. That is a selection made for consistency with the quantity being averaged. The paper discloses this and offers a plausible explanation (different membership compilations), but it remains a weak link because SAI 24 feeds into the W345 aggregate and the backward orbits. With the Tarricq value the aggregate mean shifts by roughly 25 km/s, which would change the orbit reconstructions. Three clusters also have no vLOS at all, so the aggregate averages rest on a small number of tracers.\n\nThe direct 3D regression gives 7 ± 10 km/s/kpc—consistent with no expansion but also with a mild expansion. So the \"not dispersing\" claim is really carried by the orbit integrations, which are only as good as the vLOS inputs. The abstract is a bit stronger than the data warrant; something like \"consistent with no net expansion\" would be more accurate. That said, the disagreement with Román-Zúñiga is a real result, and the projection-effect explanation is physically sensible.\n\nWho is this for? Anyone working on young star clusters, Gaia astrometry, or the Perseus / Galactic anti-center region. It deserves serious peer review, and the likely outcome is conditional acceptance after the vLOS sensitivity is tightened. I'd bring it to the reading group.","headline":"A solid, honest paper that makes a plausible case that the Perseus complex's apparent expansion is a projection effect, though the argument leans on a contested SAI 24 velocity and the abstract slightly overstates the constraint.","tokens_in":32704,"tokens_out":2096,"would_cite":true,"duration_ms":20020,"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 Perseus complex is not dispersing: the paper argues that the apparent Hubble-like expansion is a projection effect caused by stars at different line-of-sight depths orbiting the Galaxy at different speeds.","keywords":["star clusters","Perseus complex","W3/W4/W5 region","Gaia DR3","cluster kinematics","orbit integration","spiral arms","young stellar objects"],"falsifier":"Take high-resolution spectra of SAI 24 and of the three clusters with no line-of-sight velocity at all (UBC 420, Basel 10, NGC 637), and recompute the 3D velocity–distance slope and the backward orbits. If the true SAI 24 velocity comes out near +52 km/s rather than −48 km/s, or the newly measured velocities shift the W345–LISCA I convergence time by much more than the quoted few-Myr errors, the claim that the complex is not dispersing loses its support.","tokens_in":31583,"feed_emoji":"⭐","tokens_out":8887,"duration_ms":74011,"temperature":0.7,"pith_summary":"This paper argues that the Perseus complex—a large star-forming region spanning the W3/W4/W5 clouds and the LISCA I and LISCA II cluster aggregates—is not flying apart, contrary to the Hubble-like expansion flow of about 15 km/s/kpc reported for its young stars. The authors use Gaia DR3 data to recover five star clusters in the W345 region, attach six-dimensional phase-space information to clusters across the whole complex, and integrate their orbits in a Galactic potential. They find that the apparent expansion vanishes when the 1.2 kpc line-of-sight depth of the complex is accounted for: the on-sky velocity pattern is what you expect from stars orbiting the Galaxy at slightly different Galactocentric distances and speeds. The same orbits place LISCA I and W345 within a few hundred parsecs of each other 20–30 Myr ago and suggest spiral-arm perturbations can keep the pieces closer for longer. The reason to care is that it separates genuine internal dispersal of young clusters from the shearing view of a large complex seen through projection.","feed_headline":"Perseus star complex is not dispersing, 3D orbits show","feed_subtitle":"A reported 15 km/s/kpc expansion in W3/W4/W5 disappears once line-of-sight depth and Galactic orbits are included.","key_machinery":"The load-bearing tool is the star cluster used as a six-dimensional tracer: each cluster supplies a 3D position (sky position plus parallax-based distance) and a 3D velocity (proper motion plus line-of-sight velocity), so the same data can be viewed as a projected flow or as true Galactic orbits. The decisive comparison is the projected velocity–distance diagram versus the intrinsic (3D) velocity–distance diagram, computed after de-projecting cluster positions relative to IC 1805; the first shows an expansion-like slope, the second does not. The orbit integrations themselves are carried out in a Galactic potential using an action–angle-based integration scheme, with 500 resamplings of distance and line-of-sight velocity errors, and the backward integrations reverse the velocity vectors to trace where the cluster aggregates came from. For internal kinematics, the paper uses the mean radial velocity to velocity dispersion ratio $\\langle v_R\\rangle/\\sigma_R$ to quantify how strongly a young cluster is expanding.","core_discovery":"Stated on the paper's own terms: the Perseus complex is not undergoing a bulk expansion. When the clusters' on-sky proper motions are plotted against projected distance from the putative expansion center, the data reproduce the previously claimed 15 km/s/kpc (here 14 ± 6 km/s/kpc) trend; but when the same clusters are placed in three dimensions using parallaxes and line-of-sight velocities, the intrinsic velocity–distance relation is 7 ± 10 km/s/kpc, consistent with no net divergence. The complex spans about 1.2 kpc along the line of sight, so stars at different Galactic radii simply orbit at different speeds, and projection manufactures the appearance of a flow. Backward orbit integrations add a formation statement: LISCA I and the W345 system were a few hundred parsecs apart roughly 25 Myr ago, while LISCA II was born roughly 0.6–1 kpc away. Adding spiral-arm perturbations to the Galactic potential does not change the short-term picture and tends to drag clusters toward arm-density peaks, which can keep the components closer together for longer than an axisymmetric potential would.","pith_inferences":["Editorial inference: if the projection interpretation is right, other young complexes seen through a large line-of-sight depth inside spiral arms may show spurious Hubble flows; re-doing such analyses with 3D velocities would reveal how common shear artifacts are.","Editorial inference: the SAI 24 velocity disagreement is the first place to look for a revision; a decisive measurement would either confirm the adopted W345 mean or reopen the formation-scenario conclusion.","Editorial inference: the spiral-arm toy model predicts that arm trapping can keep clusters together for hundreds of Myr, which would make hierarchical merging among the Perseus components more likely; this could be tested with self-consistent N-body simulations of cluster aggregates in the same potential.","Editorial inference: the same cluster-as-6D-tracer method could be applied to other Galactic star-forming complexes to quantify the fraction of observed 'expansions' that are actually Galactic shear projection."],"forward_implications":["The 15 km/s/kpc Hubble-like expansion reported for the Perseus complex is a projection artifact; the complex itself shows no net 3D divergence (7 ± 10 km/s/kpc).","LISCA I and the W345 system were separated by only a few hundred parsecs about 20–30 Myr ago; LISCA II formed roughly 0.6–1 kpc from them.","Spiral-arm perturbations drag star clusters toward higher-density regions and can keep the components of the complex closer together for over 200 Myr compared to the axisymmetric case.","The three youngest W345 clusters (IC 1805, IC 1848, SAI 24, age ≈ 5 Myr) are internally expanding with $\\langle v_R\\rangle/\\sigma_R$ up to about 1–2, while the older Berkeley 65 and UBC 420 are near equilibrium, matching the age trend for young clusters.","Aggregating clusters into the W345, LISCA I, and LISCA II systems gives cleaner orbits than individual clusters, whose 6D initial conditions are too sensitive to line-of-sight velocity errors."],"supporting_citations":[{"why":"Supplies the claimed 15 km/s/kpc Hubble-like expansion that this paper reinterprets as a projection effect.","marker":"Román-Zúñiga et al. (2019)"},{"why":"Source of cluster positions, proper motions, distances, and the line-of-sight velocity adopted for SAI 24.","marker":"Hunt & Reffert (2023)"},{"why":"Provides line-of-sight velocities for clusters; its conflicting SAI 24 value represents the main data tension.","marker":"Tarricq et al. (2021)"},{"why":"Provides the high-resolution mean line-of-sight velocity (−39 km/s) adopted for the W345 complex, anchoring the SAI 24 choice.","marker":"Fanelli et al. (2022b)"},{"why":"Defines the axisymmetric Galactic potential used for forward and backward orbit integrations.","marker":"McMillan (2017)"},{"why":"Provides the spiral-arm geometry and pitch angles used to build the perturbed potential.","marker":"Reid et al. (2019)"},{"why":"Supplies the analytic spiral perturbation formula added to the axisymmetric potential.","marker":"Cox & Gómez (2002)"},{"why":"Gives the Perseus-arm pattern speed adopted in the spiral-perturbed orbit integrations.","marker":"Castro-Ginard et al. (2021)"},{"why":"Defines the LISCA I aggregate used as one of the three cluster complexes.","marker":"Dalessandro et al. (2021)"},{"why":"Defines LISCA II and supplies the cluster catalog and methods the present analysis builds on.","marker":"Della Croce et al. (2023)"}],"fun_headline_variants":["Perseus complex expansion is a projection illusion","No real expansion in Perseus star complex","3D orbits show Perseus complex is stable","Perseus star complex not dispersing after all","Projection tricks mimic expansion in Perseus"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The whole orbit-based story leans on the adopted line-of-sight velocities, and the most fragile single number is the velocity chosen for the cluster SAI 24, whose two catalog values (+52 ± 22 km/s and −48 ± 5 km/s) disagree by 3.7 sigma and pull the W345 mean in opposite directions.","fun_headline_variants_meta":{"raw":{"variants":["Perseus complex expansion is a projection illusion","No real expansion in Perseus star complex","3D orbits show Perseus complex is stable","Perseus star complex not dispersing after all","Projection tricks mimic expansion in Perseus"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000269,"raw_usage":{"total_tokens":1724,"prompt_tokens":1149,"completion_tokens":575,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":765,"completion_tokens_details":{"reasoning_tokens":502}},"tokens_in":765,"tokens_out":575,"duration_ms":4639,"temperature":1.0,"reasoning_tokens":502,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-16T11:13:44.455141+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Take high-resolution spectra of SAI 24 and of the three clusters with no line-of-sight velocity at all (UBC 420, Basel 10, NGC 637), and recompute the 3D velocity–distance slope and the backward orbits. If the true SAI 24 velocity comes out near +52 km/s rather than −48 km/s, or the newly measured velocities shift the W345–LISCA I convergence time by much more than the quoted few-Myr errors, the claim that the complex is not dispersing loses its support.","supporting_citations":[],"review_version":1}