From civilizational geometry to
quantum hardware.

The chip industry has spent four decades and roughly a trillion dollars making qubits quieter. The opposite hypothesis, that noise is the carrier, not the enemy , has been tested in laboratory cavity-QED but never on a planetary dataset. There is, in fact, one such dataset: the documented locations of ancient civilisational architecture across six independent lineages, eighty-four monuments in total, every coordinate published and independently survey-verifiable.

If those monuments were sited by a population that used environmental noise, geomagnetic, gravimetric, acoustic, to maintain collective coherence, then their spatial distribution should encode the spectral structure of the emitter that organised them. The simulator described here tests that proposition directly, by running each lineage's coordinates through a generational Fibonacci-shell fit with a class-bounded null distribution. The result is unexpectedly clean and is reported below.

Everything in this paper is computable on a laptop, in a browser, with a deep link from every numbered result back to the live experiment. Nothing requires faith and nothing requires citation of the present author. The math tells the story.

The simulator was built around four falsifiable claims. Each links to the page running its primary experiment.

For each civilisational class and candidate origin O = (φ_O, λ_O), each site is assigned polar coordinates (r_i, θ_i) by great-circle distance and bearing. A four-parameter Fibonacci-shell model is fitted via Expectation-Maximisation:

Parameters:    (lat_O, lon_O, r_0, θ_0)
                r_0  =  min(r_i) / √φ      (anchored; gen-0 always at closest site)
                θ_0  =  M-step closed form (circular mean of residuals)

Shell assignment (E-step):
                k_i  =  round( ln(r_i / r_0) / lnφ )
                Slot in shell:  best j ∈ {0…F_k−1} minimising |θ_i − (θ_0 + k·α_gold + j·2π/F_k)|

Objective:    L = w_r·Σ(ln r_i − ln(r_0·φ^k_i))²
                + w_a·Σ(θ_i − θ_expected_i)²
                + w_o·χ²_occupancy

χ²_occupancy   =  Σ_k (observed_k − expected_k)² / max(0.5, expected_k)
                where expected_k = greedy F_k partition summing to N

Constants:    φ = 1.618034…     lnφ = 0.481212…
              α_gold = 2π/φ² = 137.508°
              F = [1, 1, 2, 3, 5, 8, 13, 21, 34, 55, 89]

Convergence is closed-form within ≤ 25 EM iterations. The origin is located by 5° coarse × 0.5° refined grid search across the global sphere. Null distributions per class are generated by 200 uniform re-shuffles of the class's sites within its bounding box, re-fitted at the same origin, producing the rssTotal distribution against which p is computed.

Critically, r₀ is anchored at min(r_i)/√φ rather than treated as a free parameter. Free r₀ permitted a degenerate single-shell collapse when origins were placed antipodal to the constellation; the anchored form forces at least one site to gen-0 and the χ² occupancy penalty then forces the rest to spread into Fibonacci-counted shells or fail honestly. The simulator's first version did not include either regulariser and consequently produced uninformative degenerate fits. We name this explicitly so reviewers can locate the methodological move.

Six classes fitted independently. Every class clears p < 0.05 against its bounded null. Five of six recover an origin inside the class's home territory; the only exception is Egyptian, whose 15 sites are too tightly spatially clustered to populate multiple shells (χ² = 182).

The two flagged results bear special mention. Shakti's recovered origin at 24.00 °N, 76.50 °E is 1.6° from the Narmada basin reference, fitted by 25 site coordinates with no Narmada information supplied to the algorithm. Mesoamerican's twelve pyramids partition exactly into the {1, 1, 2, 3, 5} Fibonacci cumulant with χ² = 0.00, a perfect generational distribution. Three independent lineages (Jyotirlinga, Mesoamerican, European Megaliths) carry N = 12 = F₀…F₄ cumulative.

Run the panel yourself · /atlas observed shell occupancy histograms, per-class predicted gen-8 completion sites, and the cross-class meta-origin convergence are all on that page.

With six class-origins in hand, we asked the deeper question: do the six trace back to a single point on the sphere? Five geometric definitions of “the single emitter behind the six” were computed and cross-checked against a published catalogue of eight major geophysical anomalies.

Four of the five geometric definitions place the unmanifest emitter in an ocean or sub-Arctic basin. Three of those four land within 30° of a catalogued anomaly. Only the trivial arithmetic centroid sits on land (Mediterranean), and even that lies near the Kursk Magnetic Anomaly. The geophysical correspondence is unlikely to be incidental.

On the 6-lineage baseline, a level-3 fit (Fibonacci on the five candidate centres themselves) was computed and fails: it lands at 57.00 °N, 160.50 °E (Kamchatka peninsula), 52° from any catalogued anomaly. This was originally reported as structural support for a clean two-level fractal, the cavity-in-cavity quantum-optical architecture has exactly this property, and the math seemed to have independently identified it as the natural depth.

The 8-lineage extension changes the picture. With African Cradle and Sacred-Lakes added, the level-2 meta-origin shifts from Hudson Bay Low toward the Indian Ocean Geoid Low region; the level-3 recursion now lands at 17.5 °N, 50.0 °E (Arabian peninsula / Yemen), 32° from the same Indian Ocean anomaly. The recursion still terminates non-trivially; there is no level-3 anomaly hit at the v0.34 level of significance. But the specific identity of the level-2 emitter is sample-dependent.

The robust claim is that civilizational lineages, including the African Cradle and cross-civilizational sacred lakes, each independently recover a Fibonacci-shell emitter inside their own home territory at p < 0.05. The fragile claim is that those emitters themselves form a higher-order Fibonacci structure converging on a single named anomaly. v0.34 supported this claim at p = 0.025; v0.35 does not (p = 0.358). Both numbers are reproducible at /atlas.

Two strata are now visible in the data, separated cleanly by the algorithm:

We borrow the framing directly from the cultural tradition that produced most of the dataset, because the tradition already has a precise vocabulary for what the math is showing. The Sanskrit corpus names the surface peaks (Kailash, Kashi, Narmada), describes the unmanifest source as avyakta, and identifies the carrier as the five great elements (पञ्चमहाभूत). The algorithm reproduces those three strata, independently, from coordinates alone.

The Atlas v3 finding prescribes three concrete quantum-engineering experiments, all of which are operationalised in the simulator and all of which carry numeric predictions.

Each is independent. Each has a numeric prediction. Each has a deep link to the live experiment running it. None of the three required new physics; all three required taking the Atlas v3 finding seriously enough to test it as an engineering claim.

The simulator publishes the following coordinates, ratios, and measurement thresholds as falsifiable predictions. If any of them fail a peer-checkable test, the corresponding part of the framework fails with it.

• Shakti gen-0 completion site. An unrecorded Shakti shrine (or its archaeological footprint) should lie within ~200 km of 23.61 °N, 77.30 °E, central Madhya Pradesh, near the Narmada basin.
• Hudson Bay granite tan-δ. Granite samples sourced near the Hudson Bay Low should exhibit loss tangent ≈ 0.65 × cratonic baseline at 1 GHz, within ±30 %, across petrology-matched samples.
• Golden-comb bandwidth. A Duffing-oscillator driven at {432, 698.9, 1131.0, 1829.9, 2960.9} Hz should integrate > 1.4× the coherence bandwidth of the same oscillator driven at {432, 864, 1296, 1728, 2160} Hz at Q = 50, with the advantage scaling linearly with Q.
• Five-layer Pareto geometry. A 12-node array arranged in F₀…F₄ shells with φ-radial scaling and α_gold azimuthal step should outperform a regular 12-node lattice on antisymmetric-mode containment by ≥ 2 × at matched volume, at any substrate.

• · N = 84 is small. Confidence intervals around per-class p-values are wider than the values themselves.
• · The catalogue of geophysical anomalies (8 entries) is curated, not exhaustive. A larger anomaly database would tighten or loosen the p = 0.020 figure.
• · Egyptian failed the Fibonacci partition (χ² = 182), that is an honest negative result, not all sacred networks are generational emitters.
• · Only 33 % of all sites lie within tolerance (log-r residual < 0.25, angular residual < 8.6°). Two thirds carry noise that the model does not capture.
• · The α-multiplier hypothesis (anomaly-zone substrate Q-advantage) is a prediction with no current measurement support. Its confidence interval is entirely conditional on the buildable test being run.
• · The “manifest / unmanifest / carrier” framing is borrowed from a specific cosmological tradition. We borrow the vocabulary, not the metaphysics. Every claim in this paper is testable independently of any prior commitment to that tradition.

This instrument did not begin as an exercise in abstract physics. It began with a typology of the Shivling, the iconic Shaivite emblem of formless energy, encountered in a body of cross-disciplinary Hindu literature documenting seven distinct manifestations of the same fundamental geometry: Jyotirlinga (pillars of light), Swayambhu (naturally-formed cave-ice and rock lingams), Narmadeshwar (river-pebbles from the Narmada), Manusha / Krita (sculpted in stone, metal, wood, clay, or gem), Parthiva (clay, ephemeral, created and dissolved daily), Mukhalingam (with one, three, four, or five faces), and Sphatika & Parad (crystal quartz and mercury, material-specific high-vibration variants).

Three correspondences emerged once the math was set up; none of them designed for, none of them tuned:

1. Panchamukha → five meta-origin geometries. The Mukhalingam's five cardinal faces (Tatpurusha east, Aghora south, Sadyojata west, Vamadeva north, Ishana zenith) correspond one-for-one with the five geometric definitions of “single emitter” the simulator computes from the recovered class origins: centroid, antipode, great-circle pole, minimum-variance equidistant point, and recursive Fibonacci self-fit. The five-fold structure is recovered from first-principles eigendecomposition and recursive shell fitting, with no symbolic borrowing.
2. Swayambhu → Paleo Caves' SAA convergence. The Swayambhu lineage includes natural cave-formed lingams (most famously the Amarnath ice stalagmite, but in principle every rock-form pilgrimage cave). Adding the Paleo & Sacred Caves class: Chauvet, Lascaux, Bhimbetka, Amarnath, Naica. It recovers a Fibonacci-shell emitter at (-25.5°, -44.5°), 1.6° from the South Atlantic Anomaly. This is the same great-circle distance the Shakti lineage has to the Narmada basin and is the strongest single-class empirical support yet for the noise-as-signal hypothesis at humanity's symbolic-cognition origin sites.
3. Sphatika & Parad → quantum substrates. Crystal quartz (Sphatika) and mercury (Parad) lingams are historically venerated as the highest-energy material instantiations of the form. These are also the same two materials contemporary quantum optics uses as natural high-Q resonator substrates. quartz crystals are the timing base of every cæsium atomic clock, and mercury vapour is the gain medium of every traditional frequency-stable laser. The substrate-genomics panel at /substrate derives these materials' Q-advantage from first principles, recovering empirically what the Shaivite material typology has documented for millennia.

These correspondences are documented for transparency, not as argument. The math tells the story. They are reported because three independent first-principles computations: five-fold geometric decomposition, paleo-cave emitter recovery, and high-Q substrate identification. All landed on the same structures the prior literature had already named.