A Systems Reconstruction of Proportion, Hierarchy, and Structural Correspondence
This is a reproducible computational study of how proportional, hierarchical, and cross-domain relations in Robert Fludd's Utriusque Cosmi Historia (1617) can be formally represented, perturbed, and stability-tested using a method called Generative Historical Systems Analysis (GHSA).
⚠️ Warning: Computational reconstructions are not historical claims about Fludd unless independently supported by primary or secondary evidence. This repository enforces that separation as a hard architectural constraint (evidence classes E0–E3).
Fludd's Utriusque Cosmi Historia (Monochordum Mundanum, 1617, Tractatus I, Liber III, p. 90) explicitly encodes a Pythagorean proportional grammar — 4:1, 2:1, 3:2, 4:3 — across cosmological and musical domains. This project formalizes that grammar as a typed relational graph, defines candidate invariants, generates N=100 controlled variants under three mutation classes (topological θ_T, parametric θ_P, semantic θ_S), and computes a stability surface S(I | θ). It asks which structures survive perturbation, and does so without conflating:
- what Fludd actually states (E0);
- what secondary scholarship states (E1);
- what the analytical framework reconstructs (E2);
- what the computational experiment generates (E3).
Working, reproducible, tested. The experiment is implemented and run; the manuscript is regenerated from the actual measured outputs.
| Invariant | Global S(I|θ) | Classification | |-----------|---------------|----------------| | Proportionality (I2) | 0.820 | marginal | | Hierarchy (I1) | 0.560 | contextual | | Light/Matter Complementarity (I3) | 0.980 | validated within tested model family | | Central Equality Point (I4) | 0.970 | validated within tested model family | | Macrocosm/Microcosm Correspondence (I5) | 1.000 | validated within tested model family | | Musical Interval Mapping (I6) | 1.000 | validated within tested model family |
Note on integrity: The earlier draft reported I2 = 0.92. The repository's actual experiment produces 0.820 with the canonical seed (
20260829, N=100, τ=0.85). The manuscript is regenerated from the experiment; the experiment is the authority, not the draft.
GHSA is a reproducible method for extracting and perturbing structural relations from historical symbolic systems while explicitly separating textual evidence (E0), scholarly interpretation (E1), analytical reconstruction (E2), and generated inference (E3). A symbolic system is treated as a generative object, and its invariance is measured rather than assumed.
git clone <repo>
cd fludd-generative-architecture
make setup
make test # 57 tests
make experiment # generate figures, tables, report, manifest
make reproduce # verify == committed output (must report PASS)
make manuscript # build manuscript/manuscript.md from actual outputs
make report # regenerate research reportsrc/fludd_ga/— core package: evidence model, ontology, relational graph, invariants, perturbation engine, stability, scoring, failure classification, provenance, validation.evidence/— evidence registry (E0–E3), claims, citations.sources/— primary/secondary source manifest (machine-readable provenance).data/,models/— data schemas and model definitions.experiments/— per-experiment configuration and placement.tests/— unit, integration, validation (integrity gates), reproducibility.results/— generated experiment output (JSON, tables, figures, reports, manifests). Regenerated, never hand-edited.manuscript/— publication manuscript built from actual outputs.docs/— methodology, evidence taxonomy, reproducibility, experimental design, failure classification, threat model, provenance policy.scripts/— runner, reproducibility verifier, manuscript builder, report generator, setup.
Robinson, M. F. (2026). Generative Architecture in Robert Fludd's Utriusque
Cosmi Historia: A Systems Reconstruction of Proportion, Hierarchy, and
Structural Correspondence. [software/handbook] GitHub.
See CITATION.cff and manuscript/references.bib.
MIT (see LICENSE).