A Negative Assessment of the Dual-Spiral Isomorphic Hypothesis in the Voynich Manuscript

 On the Limits of Structural Coherence Under Generative Interpretation

Date: January 2026
Author: Angel Bayona
(following an inter-model experimental analysis with IA Claude, Copilot, Perplexity, and Gemini)

---

In fact: IAs don´t understand my real approch... so, maybe in future I can continue it the real one, now I´m on DSSn 

🤷‍♂️

Abstract

A previous study proposed that the Voynich Manuscript encodes a dual-spiral isomorphic structure linking botanical and anthropomorphic domains through mirrored textual architectures. The hypothesis was supported by structural regularities observed in two representative folios (f1r and f75r), including palindromic elements, repetition, gradual word transformations, and centrally positioned pivot tokens.

This article presents a negative reassessment of that hypothesis. Through controlled generative experiments, inter-model auditing, and falsification-oriented stress tests, we demonstrate that the observed coherence can be reproduced under constrained semantic frameworks independently of the Voynich corpus itself. The results indicate that the dual-spiral model, while internally consistent and symbolically compelling, does not yet satisfy the criteria required to distinguish intrinsic manuscript structure from imposed interpretive scaffolding.

We conclude that the dual-spiral hypothesis should be provisionally reclassified from structural discovery to interpretive projection, pending corpus-wide validation and bottom-up statistical independence tests.

1. Introduction

1.1 Background

The Voynich Manuscript remains one of the most persistent unresolved artifacts in historical cryptography and paleography. The prior paper, “The Dual-Spiral Isomorphic Structure of the Voynich Manuscript”, advanced a novel structural interpretation: that the manuscript encodes a non-linear, inward-outward spiral architecture mapping natural processes onto anthropomorphic or ritual ones.

This reinterpretation aimed to explain long-standing anomalies, including the Currier A/B distinction, non-overlapping vocabularies, and the apparent primacy of illustrations over text.

1.2 Purpose of This Reassessment

The present work does not aim to refute the symbolic plausibility of the dual-spiral model. Instead, it addresses a more fundamental question:

Does the observed structural coherence originate from the manuscript itself, or from the interpretive framework imposed upon it?

This distinction is critical. Structural elegance alone is not evidence of discovery.

2. Summary of the Original Hypothesis

The original article claimed evidence for:

• Dual-spiral textual progression (IDA–REGRESO)

• Isomorphic mapping between botanical and anthropomorphic domains

• Central “pivot” or “Herog” tokens (e.g., or, r, lol)

• Spiral contraction/expansion encoded through repetition and word mutation

• Functional explanation for Currier A vs B as projections, not languages

These claims were supported by detailed analyses of folios f1r and f75r.

3. Methodological Weak Points Identified

3.1 Sample Size and Representativeness

The hypothesis rests primarily on two folios, selected for their apparent contrast and clarity. While illustrative, this introduces a severe confirmation risk:

• No negative controls were analyzed.

• No random or structurally “boring” folios were tested.

• No corpus-wide baseline for expected repetition, suffix density, or positional symmetry was established.

Conclusion: The sample is insufficient to support generalization.

3.2 Structural Metrics Without Null Models

The detected features — palindromes, repetitions, suffix clustering — were not compared against:

• Randomized Voynichese

• Permuted line orders

• Equivalent-length synthetic corpora

Subsequent generative experiments demonstrated that comparable structural signatures emerge under arbitrary closed dictionaries, even when the input text is noise.

Conclusion: The metrics lack discriminative power without null distributions.

3.3 Framework Dependency and Semantic Absorption

When the dual-spiral framework was imposed on generative models (Claude), the models consistently produced:

• Coherent spiral narratives

• Meaningful pivot interpretations

• Stable dual readings

Crucially, these behaviors persisted even when:

• Word order was permuted

• Sections were swapped

• Non-Voynich corpora were substituted

This indicates semantic absorption: the framework dominates interpretation regardless of input.

Conclusion: Coherence here is a property of the framework, not evidence of manuscript encoding.

4. Failure Under Falsification Tests

A series of “lethal tests” were conceptually applied (full details omitted here):

• Permutation Test: Structure survived order scrambling.

• Noise Injection: Meaningful spirals emerged from corrupted input.

• Ablation Test: Removing the spiral assumption collapsed the interpretation entirely.

• Cross-Corpus Test: Similar structures appeared in unrelated symbolic texts.

These outcomes are incompatible with claims of intrinsic structural encoding.

5. Reinterpretation of Key Findings

5.1 Pivot Words (“or”, “r”, “lol”)

While these tokens are short, frequent, and centrally placed, such properties are statistically expected in:

• Zipfian distributions

• Highly constrained alphabets

• Agglutinative or formulaic systems

Their symbolic elevation to “Herog” status is interpretive, not compelled by data.

5.2 Spiral Geometry

The spiral model remains conceptually elegant, but:

• No unique geometric signature distinguishes it from cyclic or recursive interpretations.

• The same data support multiple incompatible geometries.

Thus, the spiral is a valid metaphor, not a demonstrated encoding.

6. What Remains Valid

This reassessment does not discard everything.

Surviving insights include:

• The Voynich text is structurally non-random.

• Local repetition and gradual mutation are real phenomena.

• Currier A/B differences are structural, not merely lexical.

• Illustrations likely play a non-decorative role.

However, these facts do not uniquely imply a dual-spiral isomorphic system.

7. Role of Illustrations: An Open but Untested Axis

The original paper correctly identified that illustrations may encode information orthogonal to the text. However:

• No independent structural analysis of images was performed.

• Textual hypotheses were retrofitted to visual motifs.

• DSSn-style operators (halo, projector, boundary) were suggested but not applied.

Conclusion: The image layer remains a promising but currently unexploited domain.

8. Epistemological Implications

This case illustrates a broader methodological risk in AI-assisted research:

High symbolic coherence can emerge from closed interpretive systems without any underlying truth claim being satisfied.

In such contexts, elegance becomes a liability unless aggressively counterbalanced by falsification.

9. Conclusion

The dual-spiral isomorphic hypothesis should, at present, be classified as:

A structurally coherent interpretive model lacking sufficient evidence of manuscript-intrinsic encoding.

It is not falsified in principle — but it is unsupported in fact.

Future work must proceed bottom-up, corpus-wide, and image-first, or the hypothesis risks becoming a self-sealing symbolic system.

Final Statement

The failure of the dual-spiral hypothesis is not a failure of insight, but of methodological separation between pattern recognition and pattern projection.

The Voynich Manuscript does not resist meaning; it resists premature closure.

In that sense, this negative result is not a dead end —
it is the first reliable map of where not to stand.