New UNNS Research · 121 Galaxies · 2026

Beyond the Rotation Curve
A Structural Atlas of Galaxies

Rotation curves tell us how galaxies move. The same data reveal something more: a reproducible structural identity. Across 121 SPARC galaxies, UNNS uncovers a dominant Stable Structure basin, 11 Weak Persistence galaxies, five internal families, and a new way to compare galaxies beyond the Radial Acceleration Relation.
121 Galaxies Perfect Repeat Recovery 11 Weak Persistence 5 Structural Families Persistence ≠ Connectivity
Instruments: STRUC-I v1.0.4 · STRUC-PERC-I v2.5.0 Domain: resolved SPARC rotation-curve galaxies (121 primary sample) Status: Canonical manuscript · 2026 Companion test: external RAR transfer boundary

What the Manuscript Reveals

A rotation curve is usually read as a single verdict: does a galaxy's observed motion match its visible mass, or not? This manuscript asks a different question. It sorts, canonicalizes, and interrogates each galaxy's complete baryonic-acceleration ladder across scale, and finds something the rotation curve alone never shows — a reproducible, multichannel, hierarchical structural identity.

Across 121 resolved SPARC galaxies, that identity separates a dominant Stable Structure basin from a sharply displaced Weak Persistence regime, resolves the Stable basin into five internal admissibility families, and shows that a galaxy's structural persistence, its gap-similarity fragmentation, and its connectivity reach are three distinct, dissociable properties — not one property viewed three ways. A companion transfer test then draws a firm boundary: this rich structural layer does not yet improve local, radius-matched dynamical prediction beyond the baseline Radial Acceleration Relation. The atlas is real. The transfer architecture that would exploit it is still to be found.

Live Instrument — Galaxy Structural Fingerprints Dashboard
Galaxy Structural Fingerprints DASHBOARD: the live atlas view — reproducibility gates, the Stable Structure / Weak Persistence split, the five-family admissibility registry, cross-chamber connectivity, structural analogues, and the external transfer boundary, all in one instrument panel.

🌀 From Rotation Curve to Structural Fingerprint

The manuscript does more than introduce another way of plotting galaxy data. It establishes a new scientific object: the multiscale structural fingerprint of a galaxy's canonicalized baryonic-acceleration ladder.

For each galaxy, the sampled values of gbar(R) are sorted, deduplicated, and interrogated across the chamber scale κ. The result is not a single statistic but a three-channel response morphology.

The Structural Fingerprint
Fg(κ) = [ρg(κ), DA,g(κ), ν̃g(κ)]

A three-channel response in which ρ(κ) tracks multiscale structural response, DA(κ) records admissibility deficit, and ν̃(κ) records normalized inversion vulnerability. Together, the three curves describe how the ladder responds as the chamber scale κ changes. The collection of these fingerprints across 121 SPARC galaxies forms a structural atlas.
Scientific pipeline showing a resolved galaxy rotation curve transformed into a sorted baryonic-acceleration ladder and passed through the UNNS chamber to produce a multichannel structural fingerprint.
Figure 1 — Structural Fingerprint Pipeline. From resolved SPARC mass models to a galaxy structural fingerprint. Each galaxy's sampled baryonic accelerations are canonicalized into a sorted ladder and interrogated across scale to produce the multichannel response Fg(κ).

The central achievement of the manuscript can be stated plainly:

The Central Achievement

Galaxy baryonic-acceleration ladders possess reproducible, organized, multichannel structural identities. That conclusion is supported by reproducibility, atlas geometry, physical alignment, and cross-chamber phenotype agreement — and is then subjected to an independent dynamical-transfer test.

🔁 A Genuine Structural Identity Coordinate

The strongest foundational result is the unchanged-preparation reproducibility test. Across five independent STRUC-I executions of the same 121-galaxy corpus, each galaxy returns to its own structural-response location and preserves its local neighbourhood in the atlas — not an unstable numerical artifact, but a stable coordinate.

Top-1 source recovery
1.000
across 5 STRUC-I repeats
Top-5 source recovery
1.000
full neighbourhood preserved
Median ICC
0.99985
intraclass correlation, ρ-PC1–3
Top-5 Jaccard overlap
1.000
median across galaxies

Why This Matters

Each galaxy's own repeated response lies far closer than the nearest incorrect galaxy's response. The fingerprint behaves as a genuine structural identity coordinate under fixed observation and preparation — the precondition for everything that follows.

🧭 An Organized Atlas, Not an Unstructured Cloud

The fingerprints do not scatter randomly. The three response channels have markedly different internal geometries — a division of structural labour between a compressible, low-dimensional signal and a richer, more heterogeneous one.

Variance Captured in Three Principal Components, by Channel ρ(κ) 90.8% ν̃(κ) 88.5% D_A(κ) 55.9% combined 61.0% ρ and ν̃ are strongly compressible; D_A is richer and less reducible to a compact geometry Variance explained by the first three principal components per channel · 121-galaxy atlas

The atlas is therefore multichannel in a substantive sense: the channels are not interchangeable measurements of one underlying scalar. They expose different dimensions of galaxy structure.

⚖️ Two Regimes: Stable Structure and Weak Persistence

The 121-galaxy sample separates into 110 Stable Structure galaxies and 11 Weak Persistence galaxies. This is not a soft or arbitrary label — the Weak Persistence population is strongly displaced in response space.

Structural-response atlas showing a dense population of 110 Stable Structure galaxies and a displaced group of 11 Weak Persistence galaxies.
Figure 2 — Stable Structure and Weak Persistence. The structural atlas separates a dominant Stable Structure basin from a distinct Weak Persistence regime. The separation is especially pronounced along the leading ρ-response direction.
Stable Structure
110
median ρ-PC1 = −2.11
Weak Persistence
11
median ρ-PC1 = +11.22

The eleven validated Weak Persistence members: ESO563-G021, F574-1, IC2574, NGC1003, NGC2998, NGC3198, NGC3972, NGC5985, NGC7793, UGC02487, and UGC11820.

The Atlas's First Large-Scale Division

A dominant stable basin and a clearly displaced weak-persistence direction — the first large-scale structural division in the atlas, and the foundation on which the finer admissibility taxonomy below is built.

🧬 Five Admissibility Families Inside Stable Structure

The Stable Structure population is itself not uniform. Using Ward clustering, Gaussian-mixture modelling, HDBSCAN, bootstrap subsampling, and consensus co-assignment, the analysis resolves a five-family admissibility structure within the 110 Stable Structure galaxies.

Consensus cluster map of the Stable Structure population showing five admissibility families, their relative sizes, and the strong Full Percolation enrichment of Cluster 2.
Figure 3 — Five Admissibility Families. Within the Stable Structure basin, consensus clustering resolves five admissibility families. Their internal co-assignment is strong, and the families differ significantly in conventional galaxy properties and connectivity behaviour.
FamilyMembersFull PercolationCharacter
Cluster 0654.6%Bright, high-Vflat, high-surface-brightness core
Cluster 11323.1%Late-type, low-luminosity, low-surface-brightness family
Cluster 21844.4%Intermediate physical scale; strongly enriched in Full Percolation
Cluster 370.0%Small family; uniformly Hard Fragmentation
Cluster 4714.3%Most ρ-PC1-displaced Stable Structure family

Internal co-assignment coherence is strong, ranging from 0.819 to 0.967, while between-family co-assignment stays low; the clustering also survives resampling with moderate-to-good agreement. More importantly, the families differ in properties that were never used to construct them: Hubble type, flat rotation velocity, 3.6 μm luminosity, effective surface brightness, H I mass, ladder-gap geometry, and STRUC-PERC-I connectivity behaviour.

The Atlas Hierarchy
Atlas ⊃ Structural regimes ⊃ Admissibility families ⊃ Structural neighbours
Nested Structural Hierarchy — 121-Galaxy Atlas ATLAS · 121 galaxies STRUCTURAL REGIMES · Stable (110) / Weak Persistence (11) ADMISSIBILITY FAMILIES · 5 consensus clusters (65 / 13 / 18 / 7 / 7) STRUCTURAL NEIGHBOURS · pairwise response similarity A reusable structural taxonomy, not a single visualization

🔗 Persistence, Fragmentation, and Connectivity Are Different

This may be the manuscript's deepest conceptual result. Across the sample, 110 galaxies show Stable Structure, 103 show Hard Fragmentation, 94 are simultaneously Stable Structure and Hard Fragmentation, and only 17 reach Full Percolation — all 17 requiring an adaptive extension beyond the native κ range.

Cross-chamber comparison showing that most galaxies are structurally persistent yet fragmented, while full percolation is uncommon and unevenly distributed among admissibility clusters.
Figure 4 — Persistence, Fragmentation, and Connectivity. Persistence, fragmentation, and connectivity are distinct structural dimensions. A galaxy may preserve a stable multiscale response while remaining fragmented and reaching full percolation only through adaptive extension.
Cross-Chamber Counts (n = 121) Stable Structure 110 / 121 Hard Fragmentation 103 / 121 both 94 / 121 Full Percolation: 17/121 (14.0%)
A Three-Coordinate Structural Description
Sg = (Pg, Φg, Cg)

where Pg is persistence, Φg is fragmentation behaviour, and Cg is connectivity reach. A structure can be persistent without being connected. Fragmentation does not imply structural failure. A galaxy ladder may preserve a stable multiscale response while remaining fragmented in its gap-similarity graph — requiring a larger connectivity scale to percolate, yet retaining strong admissibility persistence throughout the native chamber interval.

A Cross-Chamber Phenotype

The five admissibility families were derived entirely from STRUC-I response geometry — yet one family is independently distinguished by STRUC-PERC-I connectivity. Cluster 2 contains 8/18 = 44.4% Full Percolation galaxies, compared with 7/92 = 7.6% across the remaining Stable Structure families.

Fisher exact test (one-sided)
p = 3.6×10⁻⁴
Cluster 2 vs. remaining Stable families
Odds ratio
≈ 9.7
Cluster 2 connectivity enrichment

Two Independent Chambers, One Subgroup

The admissibility families are not only geometrical partitions of one PCA plot. At least one family possesses a distinct cross-chamber connectivity phenotype — STRUC-I structure geometry and STRUC-PERC-I connectivity, measured independently, converge on the same galaxies.

🧩 Physical Organization Without Reducing to Classification

Structural-PC2 correlates with effective surface brightness, 3.6 μm luminosity, flat rotation velocity, and H I mass. Structural nearest neighbours are more physically alike than the all-galaxy-pair baseline — but the atlas also reveals relationships conventional classification does not directly capture.

Hubble type
−33%
median neighbour difference
Flat velocity
−22%
vs. all-pair baseline
Luminosity
−43%
3.6 μm, vs. baseline
Surface brightness
−31%
vs. baseline
H I mass
−37%
vs. baseline

Structural Analogues — A New Relation Between Galaxies

Pairs such as DDO 168 – NGC 3893 and NGC 5585 – NGC 7814 are structurally close despite large differences in conventional morphology and velocity scale. These are structural analogues: galaxies occupying similar multiscale response positions without belonging to the same traditional class. The atlas introduces a relation — similarity of admissibility-response morphology — that is not equivalent to Hubble type, luminosity, rotation speed, or surface brightness.

🔬 How Observation Transforms a Fingerprint

Gate C shows that the recorded fingerprint changes systematically when the observational representation changes. Single-radius deletion and mass-to-light variation produce moderate transformations; deterministic thinning and radial-coverage changes produce much stronger movement.

The Observational Transformation Layer
Fgobs = 𝒯obs[Fggal]

This does not say the fingerprint is unreal. It says structural identity is expressed through an observational channel, and that changes in coverage, point density, ladder gaps, and baryonic preparation transform the observed coordinates.

The Concrete Next Step — A Normalization Law
g = 𝒩(Fgobs, Og)

where Og records observational geometry and 𝒩 seeks to recover a normalized structural identity. The project has moved from merely detecting fingerprints to beginning to understand their transformation law.

🌌 Where This Sits Relative to the RAR, MOND, and ΛCDM

The Radial Acceleration Relation and the UNNS atlas describe different layers of galaxy organization.

The RAR — Local Relation

gobs(R) ≈ F(gbar(R)). It links baryonic acceleration and observed dynamical acceleration at matched radius — a local coupling.

The UNNS Atlas — Whole-Ladder Structure

It describes the global multiscale structure of the complete canonicalized baryonic-acceleration ladder — a whole-ladder structural identity, not a radius-matched relation.

A companion test — the Gravity Bridge — asked whether a static galaxy-level fingerprint could simply be added to a local RAR-like model to improve held-out prediction. It could not. The best model remained the local baseline using gbar and normalized radius; compact chamber summaries worsened held-out transfer, and even whole-galaxy ladder statistics did not improve the local model.

An Architectural Boundary, Not a Failure

This does not invalidate the atlas — it identifies the wrong transfer architecture. The rejected assumption is that one complete galaxy fingerprint can be inserted as one static global descriptor into one universal radial prediction law. The data instead point toward local, regime-conditioned, transition-sensitive, and channel-selective models.

The ρ(κ) Channel: The Most Promising Lead

The full ρ(κ) curve emerged as the leading transfer-sensitive channel: it improved aggregate prediction relative to M2, the registered whole-galaxy comparison baseline, although M1 remained the best-performing model overall. It also improved four of five cross-validation folds, improved predictions for 71 of 121 galaxies, outperformed the all-channel combination, and remained useful where the ν̃ channel was detrimental. It did not clear the manuscript's complete registered confirmation gate — but it marks the most promising route forward.

TheoryWhat the Atlas Adds
MONDRaises a new question: why do galaxies following a common acceleration law occupy distinct persistence regimes, admissibility families, fragmentation states, and connectivity classes?
ΛCDMSupplies a new empirical target — a simulation reproducing the mean RAR can now also be asked whether it reproduces the regime split, the five-family organization, and the cross-class structural analogues.

Theory-Neutral, Theory-Demanding

The atlas does not attempt to prove or disprove MOND, nor does it argue for or against ΛCDM. It gives both modified-dynamics and dark-matter models a richer set of structures to explain than the RAR alone.

🚀 Broader Implications

The atlas opens several practical and theoretical directions for galaxy science and for the UNNS programme itself.

Classification

Structural Comparison

Galaxies compared by multiscale structural response rather than only morphology, mass, or rotation speed.

New relation established
Anomaly Detection

Weak Persistence & Outliers

Weak Persistence objects, structural outliers, and unusual cross-chamber combinations can be targeted for closer study.

11 flagged already
Survey Comparison

Normalized Fingerprints

Normalized fingerprints could allow structural comparison across surveys with different radial coverage and sampling density.

Motivated by Gate C
Simulation Testing

Synthetic Galaxy Projection

Synthetic galaxies can be projected into the same atlas and tested for regime, cluster, and connectivity agreement.

Ready-made empirical target
Observation Planning

Minimum Coverage

Gate C reveals which aspects of radial sampling most strongly affect fingerprint recovery.

Practical design input
Regime-Conditioned Dynamics

Family-Specific Models

Future gravity or acceleration models can be trained separately within admissibility families rather than one coefficient for all galaxies.

Direct manuscript path

📖 The Manuscript and Its Principal Findings

Editorial illustration of the Galaxy Structural Fingerprints manuscript beside a summary of its principal validated findings, including reproducibility, structural regimes, admissibility families, and transfer limits.
Figure 5 — Manuscript and Principal Findings. Galaxy Structural Fingerprints presents a reproducible multiscale atlas of 121 baryonic-acceleration ladders, revealing structural identity, a distinct Weak Persistence regime, five admissibility families, and separate persistence, fragmentation, and connectivity dimensions.

For the UNNS programme, this project demonstrates that its chambers can produce highly reproducible object-specific identities, a genuine multichannel structural coordinate system, discrete regimes, nested admissibility families, cross-chamber phenotype alignment, meaningful relations to independent physical variables, and explicit boundaries between structural identity and predictive transfer. The result is a concrete, reusable research object: the galaxy structural-fingerprint atlas.

💡 The Main Revelation

The most important revelation is not simply that galaxies can be clustered. It is that a galaxy's baryonic organization contains a structural layer that is reproducible, multiscale, hierarchical, partly physical, partly observationally transformed, and distinct from local acceleration-law prediction.

Final Synthesis
A galaxy can obey the same broad Radial Acceleration Relation as another galaxy while occupying a different persistence regime, admissibility family, fragmentation state, connectivity reach, and structural neighbourhood. The local acceleration relation does not exhaust the organization present in the baryonic data. The failed static transfer model marks an architectural boundary, not the absence of structural information — and the resulting atlas is a new empirical object for UNNS, galaxy dynamics, modified-gravity theories, dark-matter models, and future observational and simulation studies.

Resources & References

UNNS Substrate Research Program · Galaxy Structural Fingerprints · 2026 · Instruments: STRUC-I v1.0.4, STRUC-PERC-I v2.5.0 · 121 SPARC galaxies · Top-1/top-5 recovery 1.000 · Five-family consensus admissibility structure · All data available for independent verification · unns.tech