pith. machine review for the scientific record. sign in

arxiv: 2604.18921 · v1 · submitted 2026-04-20 · 🌌 astro-ph.SR · astro-ph.GA· astro-ph.IM

Recognition: unknown

Stellar separation shapes spin-orbit alignment in visual binaries

Michael Poon , Dang Pham , Marta L. Bryan , Hanno Rein , Jiayin Dong
Authors on Pith no claims yet

Pith reviewed 2026-05-10 03:04 UTC · model grok-4.3

classification 🌌 astro-ph.SR astro-ph.GAastro-ph.IM
keywords visual binariesspin-orbit alignmentstellar separationbinary formationhierarchical Bayesian modelFisher distributionprotostellar disk
0
0 comments X

The pith

Visual binary stars show stronger spin-orbit alignment when separated by less than about 35 AU.

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

Visual binaries are pairs of stars whose orbits and spins can be measured from Earth. The degree of alignment between a star's spin axis and the binary orbit plane may encode how the pair formed, either from a shared disk or through turbulent fragmentation of a cloud. Previous analyses were hampered by inaccurate stellar radii, but this study applies a hierarchical Bayesian statistical model to corrected data. It finds robust evidence for two distinct populations split by a separation of roughly 31 to 38 AU: closer binaries are highly aligned, while wider ones show weaker alignment. This supports the idea that separation acts as a marker between different binary formation channels.

Core claim

Using a hierarchical Bayesian model on updated spin-orbit angle measurements for visual binaries, the analysis yields a Bayes factor of 12 in favor of two subpopulations divided by a cutoff at approximately 31-38 AU. Binaries interior to this cutoff are consistent with a Fisher distribution of high concentration parameter κ=48, indicating strong alignment, whereas exterior binaries follow a distribution with κ=6, indicating weaker alignment. Indications of a possible secondary cutoff around 10-17 AU are also present but require additional observations to confirm. These results suggest transitions between formation pathways, with closer binaries forming aligned within a protostellar disk and

What carries the argument

Hierarchical Bayesian model fitting two subpopulations of spin-orbit angles separated at a stellar separation cutoff, modeled with Fisher distributions of differing concentration parameters κ.

If this is right

  • Closer binaries below the cutoff form through disk fragmentation and inherit aligned spins and orbits.
  • Wider binaries above the cutoff form via turbulent fragmentation and exhibit more random orientations.
  • A secondary transition may exist at even smaller separations around 10-17 AU.
  • Accurate stellar radius corrections are essential for reliable population-level inferences on alignment.

Where Pith is reading between the lines

These are editorial extensions of the paper, not claims the author makes directly.

  • Future observations targeting binaries near the 30-40 AU boundary could sharpen the location and sharpness of the cutoff.
  • The model might be extended to include other binary parameters such as eccentricity or mass ratio to test for correlated effects.
  • Similar separation-dependent alignments could be searched for in wider populations like triple systems or in different stellar environments.

Load-bearing premise

The corrected spin-orbit angles in the dataset accurately reflect true values and the Bayesian model can reliably identify a physical cutoff without being misled by data selection biases or statistical artifacts.

What would settle it

A new sample of visual binaries with separations near 35 AU that shows no difference in alignment statistics between the close and wide groups would falsify the two-subpopulation claim.

Figures

Figures reproduced from arXiv: 2604.18921 by Dang Pham, Hanno Rein, Jiayin Dong, Marta L. Bryan, Michael Poon.

Figure 1
Figure 1. Figure 1: Fisher distribution for several values of κ, show￾ing the resulting spin-orbit angle ψ probability density func￾tions. Small κ values produce nearly isotropic distributions, whereas large κ values indicate strong alignment [PITH_FULL_IMAGE:figures/full_fig_p002_1.png] view at source ↗
Figure 2
Figure 2. Figure 2: Graphical representation of our two hierarchical models. Arrows represent the direction of data generation either from the Single Population Model or the Piecewise Model to ψi, the spin-orbit angle of star i, to the observed variables ˆis (stellar spin axis inclination) and ˆio (orbital in￾clination). that some binaries < 20 AU in the A. Hale (1994) in￾stead have greater misalignment. The JA20 dataset (cf.… view at source ↗
Figure 4
Figure 4. Figure 4: Posterior distribution of the semi-major axis cut￾off (acutoff ) for the piecewise model, where vertical dotted lines represent the semi-major axis of each stellar binary. The prior is a uniform distribution. The region with the highest probability is acutoff ∼ 31 − 38 AU. of κ = 11+10 −6 . This indicates that the distribution of spin-orbit angles for a single population have a peak at ψ ∼ 17◦ (bottom pane… view at source ↗
Figure 3
Figure 3. Figure 3: Top: Posterior distribution of κ for the Sin￾gle Population Model. The mode and 68% highest density probability interval is κ = 11+10 −6 , corresponding to spin-orbit angle distributions peaking at ψ = 17+8 −5 ◦ . Bottom: 1000 spin-orbit angle distributions shown as blue curves, each cor￾responding to a sample from the κ posterior. In black is the ψ distribution for κ = 11 which peaks at 17◦ , signifying m… view at source ↗
Figure 5
Figure 5. Figure 5: The Bayes factor (BFpiecewise, single pop.) compar￾ing the piecewise and single population models. The evidence for the piecewise is evaluated as a function of semi-major axis cutoff (acutoff ), where vertical dotted lines represent the semi-major axis of each stellar binary. Horizontal dashed lines represent the Bayes factor interpretation by H. Jeffreys (1939). A Bayes factor greater than 1 favors the pi… view at source ↗
Figure 6
Figure 6. Figure 6: Posteriors for the two population model. Top left: The posterior for κ1 (subpopulation with semi-major axis a ≤ acutoff ). The mode and 68% highest density probability interval (HDPI) is κ1 = 48+194 −32 , corresponding to spin-orbit angles peaking at ψ = 8+6 −4 ◦ . Top right: 1000 ψ distributions shown as green curves, each corresponding to samples from the κ1 posterior. The black curve is the ψ distributi… view at source ↗
Figure 7
Figure 7. Figure 7: Projected spin-orbit angle |is−io| for each star in the JA20 dataset versus semi-major axis. In blue, the violin plot shows the probability distribution for each system, with black points indicating the median. The vertical black line separates the two populations at acutoff = 31 − 38 AU. Inte￾rior (exterior) to acutoff , the green (orange) dotted line rep￾resents the 3σ bound of possible projected spin-or… view at source ↗
Figure 8
Figure 8. Figure 8: The Bayes factor for the piecewise (two populations) and single population model. Each curve corresponds to different assumed orbital inclination errors (1◦ , 5 ◦ , 10◦ ) to test the sensitivity of our Bayes factor calculations. The data in JA20 assumed no orbital inclination error (black curve, same as [PITH_FULL_IMAGE:figures/full_fig_p009_8.png] view at source ↗
Figure 9
Figure 9. Figure 9: Posterior distributions of κ for the < 30 AU and > 60 AU subsamples, showing peak values consistent with those inferred from the two-population piecewise model. where the Fisher distribution is PFisher(κ) = κ 2 sinh κ exp (κ cos ψ) sin ψ. (B2) Like in the piecewise model, κ1 and κ2 describe the two populations’ spin-orbit angle distribution. The variable f describes the two populations’ weighting. Note tha… view at source ↗
Figure 10
Figure 10. Figure 10: Posterior distributions of κ1, κ2, and f for the mixture model. For the κ posteriors, we also show the 1000 corresponding ψ distributions on the right panel. The mode and 68% highest density probability interval (HDPI) are κ1 = 26+179 −19 , κ2 = 5+5 −3 , and f = 0.66+0.21 −0.29. This result indicates that there are two distinct stellar spin-orbit angle distributions. 1 = 48 +194 32 80 160 240 320 400 1(a … view at source ↗
Figure 11
Figure 11. Figure 11: Corner plot for the two populations of spin-orbit angle (piecewise) model. The values shown in the figure are the mode and 68% highest density probability interval [PITH_FULL_IMAGE:figures/full_fig_p011_11.png] view at source ↗
Figure 12
Figure 12. Figure 12: Corner plot for the three populations of spin-orbit angle model. The values shown in the figure are the mode and 68% highest density probability interval [PITH_FULL_IMAGE:figures/full_fig_p012_12.png] view at source ↗
read the original abstract

Stellar binaries may form through several formation pathways, including disk or core fragmentation. Their spin-orbit angles are a signature of formation, although individual measurements for visual binaries are limited and broad. A seminal work by A. Hale (1994) found that visual binaries with separations $\lesssim 30$ AU tend to be more aligned, which laid the groundwork for binary formation theories. However, A. B. Justesen & S. Albrecht (2020) found that underestimated stellar radii lead to inaccurate spin-orbit angles and that KS statistics do not provide meaningful population-level constraints even with updated radii. Using a hierarchical Bayesian model to reanalyze their dataset, we find evidence with a Bayes factor of 12 for two subpopulations of spin-orbit angles separated by a $\sim 31-38$ AU cutoff. Binaries inside (outside) the cutoff are more (less) aligned, consistent with a Fisher distribution with $\kappa=48$ ($\kappa=6$). We also find possible indications of a secondary cutoff at $\sim 10-17$ AU, although more data is required to resolve this prediction. These cutoffs may mark transitions between formation pathways: closer-in binaries tend to form aligned in a shared protostellar disk, while wider binaries tend to form less aligned through turbulent fragmentation.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, simulated authors' rebuttal, and a circularity audit. Tearing a paper down is the easy half of reading it; the pith above is the substance, this is the friction.

Referee Report

2 major / 2 minor

Summary. The manuscript reanalyzes the Justesen & Albrecht (2020) dataset of visual binaries using a hierarchical Bayesian model. It reports a Bayes factor of 12 favoring two subpopulations of spin-orbit angles partitioned at a fitted primary cutoff of ~31-38 AU, with closer systems consistent with a Fisher distribution of concentration κ=48 and wider systems with κ=6; a possible secondary cutoff at ~10-17 AU is also noted. These results are interpreted as evidence for distinct formation pathways (disk vs. turbulent fragmentation) depending on separation.

Significance. If the statistical inference is robust, the work provides the first quantitative population-level constraints on separation-dependent alignment in visual binaries, moving beyond the inconclusive KS tests in prior work and offering falsifiable predictions for formation models. The use of a hierarchical Bayesian framework with explicit subpopulation modeling is a methodological strength.

major comments (2)
  1. [Abstract and methods] The abstract states a Bayes factor of 12 but provides no information on the model likelihood, priors on the cutoff and κ parameters, data exclusion criteria, or propagation of radius uncertainties into the spin-orbit angles. Without these details (presumably in the methods or §3), it is impossible to assess whether the reported evidence is sensitive to modeling choices or data selection.
  2. [Results on subpopulation inference] The subpopulations are defined by the fitted 31-38 AU cutoff, after which the κ values are inferred from the partitioned data. This introduces a risk of circular dependence between the separation threshold and the alignment parameters; the manuscript should include posterior predictive checks or simulation-based validation to demonstrate that the Bayes factor is not inflated by this partitioning procedure.
minor comments (2)
  1. [Introduction] The reference to Hale (1994) should include a brief recap of the original sample size and separation range to allow direct comparison with the current dataset.
  2. [Results] Notation for the Fisher concentration parameter κ should be defined explicitly on first use, and the secondary cutoff at 10-17 AU should be accompanied by its own Bayes factor or credible interval for clarity.

Simulated Author's Rebuttal

2 responses · 0 unresolved

We thank the referee for their insightful comments on our manuscript. We address each of the major comments below and have revised the manuscript accordingly where necessary to enhance clarity and robustness.

read point-by-point responses

  1. Referee: [Abstract and methods] The abstract states a Bayes factor of 12 but provides no information on the model likelihood, priors on the cutoff and κ parameters, data exclusion criteria, or propagation of radius uncertainties into the spin-orbit angles. Without these details (presumably in the methods or §3), it is impossible to assess whether the reported evidence is sensitive to modeling choices or data selection.

    Authors: We agree that the abstract, being a concise summary, does not include all methodological details. The full description of the hierarchical Bayesian model, including the likelihood function, the priors on the cutoff and κ parameters, the data selection criteria, and the propagation of radius uncertainties into the spin-orbit angles are all provided in detail in Section 3 (Methods) of the manuscript. To improve accessibility, we will add a brief sentence to the abstract summarizing the key modeling aspects. We have also ensured that the methods section explicitly references these elements for the reader's convenience. revision: partial

  2. Referee: [Results on subpopulation inference] The subpopulations are defined by the fitted 31-38 AU cutoff, after which the κ values are inferred from the partitioned data. This introduces a risk of circular dependence between the separation threshold and the alignment parameters; the manuscript should include posterior predictive checks or simulation-based validation to demonstrate that the Bayes factor is not inflated by this partitioning procedure.

    Authors: We appreciate this concern regarding potential circularity. In our model, the cutoff separation is a free parameter that is inferred jointly with the κ values for the two subpopulations in a single hierarchical Bayesian framework. The model likelihood accounts for the probability of each binary belonging to either subpopulation based on its separation relative to the cutoff. The Bayes factor is computed by comparing the evidence for this two-component model against a single-component model. To directly address the referee's suggestion, we have conducted posterior predictive checks: we simulated new datasets from the posterior distribution of the parameters and verified that the model recovers the input cutoff and κ values without bias. These validation results will be included in the revised manuscript as a new subsection or figure. revision: yes

Circularity Check

0 steps flagged

No significant circularity detected

full rationale

The paper applies a hierarchical Bayesian model to an existing dataset, inferring a separation cutoff and Fisher-distribution parameters (kappa) for subpopulations as jointly estimated model parameters. The abstract and description present this as standard population inference yielding a Bayes factor comparison, with no equations, self-citations, or steps shown that reduce the result to the inputs by construction. No load-bearing self-citation, uniqueness theorem, ansatz smuggling, or renaming of known results is present. The derivation chain is self-contained statistical modeling on the data.

Axiom & Free-Parameter Ledger

4 free parameters · 2 axioms · 1 invented entities

The central claim rests on fitted separation cutoffs and concentration parameters derived from a hierarchical Bayesian model applied to reanalyzed observational data; the model assumes Fisher distributions for each subpopulation and treats the input angles as reliable after prior radius corrections.

free parameters (4)
  • primary cutoff separation = 31-38 AU
    Fitted threshold that partitions the sample into more-aligned and less-aligned subpopulations
  • kappa close = 48
    Concentration parameter of the Fisher distribution for the close subpopulation
  • kappa wide = 6
    Concentration parameter of the Fisher distribution for the wide subpopulation
  • secondary cutoff = 10-17 AU
    Possible additional transition point suggested by the data
axioms (2)
  • domain assumption Spin-orbit angles follow a Fisher distribution within each subpopulation
    Used to model the strength of alignment for close versus wide binaries
  • domain assumption The reanalyzed dataset provides accurate spin-orbit angles after stellar radius corrections
    The analysis builds directly on the updated measurements from Justesen & Albrecht (2020)
invented entities (1)
  • Two distinct subpopulations of spin-orbit alignments no independent evidence
    purpose: To account for the observed change in alignment strength across the separation cutoff
    Inferred from the Bayesian model fit rather than directly observed

pith-pipeline@v0.9.0 · 5540 in / 1789 out tokens · 59606 ms · 2026-05-10T03:04:24.126961+00:00 · methodology

discussion (0)

Sign in with ORCID, Apple, or X to comment. Anyone can read and Pith papers without signing in.

Reference graph

Works this paper leans on

18 extracted references · 17 canonical work pages

  1. [1]

    Bate, M. R. 2009, MNRAS, 392, 590, doi: 10.1111/j.1365-2966.2008.14106.x

    work page doi:10.1111/j.1365-2966.2008.14106.x 2009

  2. [2]

    Bate, M. R. 2012, MNRAS, 419, 3115, doi: 10.1111/j.1365-2966.2011.19955.x

    work page doi:10.1111/j.1365-2966.2011.19955.x 2012

  3. [3]

    Bate, M. R. 2018, MNRAS, 475, 5618, doi: 10.1093/mnras/sty169

    work page doi:10.1093/mnras/sty169 2018

  4. [4]

    1953, Proceedings of the Royal Society of London Series A, 217, 295, doi: 10.1098/rspa.1953.0064

    Fisher, R. 1953, Proceedings of the Royal Society of London Series A, 217, 295, doi: 10.1098/rspa.1953.0064

    work page doi:10.1098/rspa.1953.0064 1953

  5. [5]

    and Lang, Dustin and Goodman, Jonathan , title =

    Foreman-Mackey, D., Hogg, D. W., Lang, D., & Goodman, J. 2013, PASP, 125, 306, doi: 10.1086/670067

    work page doi:10.1086/670067 2013

  6. [6]

    1994, AJ, 107, 306, doi: 10.1086/116855

    Hale, A. 1994, AJ, 107, 306, doi: 10.1086/116855

    work page doi:10.1086/116855 1994

  7. [7]

    1939, Theory of Probability

    Jeffreys, H. 1939, Theory of Probability

    1939

  8. [8]

    B., & Albrecht, S

    Justesen, A. B., & Albrecht, S. 2020, A&A, 642, A212, doi: 10.1051/0004-6361/202039138

    work page doi:10.1051/0004-6361/202039138 2020

  9. [9]

    URLhttps://doi.org/ 10.1080/01621459.1995.10476572

    Kass, R. E., & Raftery, A. E. 1995, J. Am. Statist. Assoc., 90, 773, doi: 10.1080/01621459.1995.10476572

    work page doi:10.1080/01621459.1995.10476572 1995

  10. [10]

    L., & Albrecht, S

    Marcussen, M. L., & Albrecht, S. H. 2022, ApJ, 933, 227, doi: 10.3847/1538-4357/ac75c2

    work page doi:10.3847/1538-4357/ac75c2 2022

  11. [11]

    L., Albrecht, S

    Marcussen, M. L., Albrecht, S. H., Winn, J. N., et al. 2024, ApJ, 975, 149, doi: 10.3847/1538-4357/ad75fa

    work page doi:10.3847/1538-4357/ad75fa 2024

  12. [12]

    Masuda, K., & Winn, J. N. 2020, AJ, 159, 81, doi: 10.3847/1538-3881/ab65be

    work page doi:10.3847/1538-3881/ab65be 2020

  13. [13]

    Mathieu, R. D. 1994, ARA&A, 32, 465, doi: 10.1146/annurev.aa.32.090194.002341 Mu˜ noz, D. J., & Perets, H. B. 2018, AJ, 156, 253, doi: 10.3847/1538-3881/aae7d0

    work page doi:10.1146/annurev.aa.32.090194.002341 1994

  14. [14]

    Fielding, D. B. 2016, ApJL, 827, L11, doi: 10.3847/2041-8205/827/1/L11

    work page doi:10.3847/2041-8205/827/1/l11 2016

  15. [15]

    L., Rein, H., et al

    Poon, M., Bryan, M. L., Rein, H., et al. 2025, ApJL, 994, L48, doi: 10.3847/2041-8213/ae1f0e

    work page doi:10.3847/2041-8213/ae1f0e 2025

  16. [16]

    L., Rein, H., et al

    Poon, M., Bryan, M. L., Rein, H., et al. 2024, AJ, 168, 270, doi: 10.3847/1538-3881/ad84e5

    work page doi:10.3847/1538-3881/ad84e5 2024

  17. [17]

    L., Moe, M., & Kratter, K

    Smith, C. L., Moe, M., & Kratter, K. M. 2024, ApJ, 975, 153, doi: 10.3847/1538-4357/ad6dd2

    work page doi:10.3847/1538-4357/ad6dd2 2024

  18. [18]

    Speagle, J. S. 2020, MNRAS, 493, 3132, doi: 10.1093/mnras/staa278 Spin-orbit alignment in visual binaries9 101 102 acutoff [AU] 100 101 BFpiecewise, single pop. Piecewise model preferred Single pop. model preferred Substantial evidence Strong evidence No io error 1.0 5.0 10.0 Figure 8.The Bayes factor for the piecewise (two populations) and single populat...

    work page doi:10.1093/mnras/staa278 2020