Theatre Orientation

Research question

Do ancient theatre orientations and theatre-to-theatre spatial relationships show a statistically demonstrable organising pattern, or are apparent directional correspondences compatible with chance, catalogue structure and a small number of influential records?

A second, narrower question was subsequently tested: among all eligible ordered theatre triplets, how unusual is the combined Epidaurus–Corinth–Delphi configuration when orientation agreement, intermediate-point geometry and cross-track displacement are considered together?

Dataset and orientation model

The present catalogue contains 58 ancient theatre records. Theatre orientation is treated directionally: the principal audience-facing azimuth runs from the seating area towards the orchestra and stage direction. Axial reduction to 0–180° is used only for parallelism and grouping, not for theatre-to-theatre targeting.

The dataset distinguishes exact theatre-monument coordinates from broader site-representative coordinates and records orientation-source quality so that lower-precision records can be isolated in sensitivity analyses.

General inter-theatre test

Run 01B evaluated 3,304 directed theatre-to-theatre comparisons at angular thresholds of 0.5°, 1°, 2° and 3°. It used GeographicLib WGS84 geodesics, distance and regional strata, source-quality and coordinate-scope strata, leave-one-source-out and leave-one-target-out influence tests, and dedicated axial analysis.

Three null models were tested with 50,000 Monte Carlo simulations each: independent orientation rotations, permutation of the observed orientation set and within-region permutation. Benjamini–Hochberg false-discovery-rate correction was applied across the declared test family.

General result: no demonstrated network

The observed totals were modestly above several null-model means, but the lowest global raw p-value was 0.0868 and its corrected q-value was 0.5141. There were no reciprocal theatre pairs within 3°.

The apparent excess was also concentrated disproportionately in lower-precision site-representative source points and a small number of influential theatres, rather than in the exact theatre-coordinate subset.

This is a useful negative result: it constrains interpretation and prevents isolated correspondences from being presented as evidence for a universal theatre grid.

Epidaurus–Corinth–Delphi configuration

Two focused Run 01B observations survived as individual cases: the Great Theatre of Epidaurus points within approximately 0.67° of the bearing to Corinth Theatre and within approximately 0.81° of the bearing to Delphi Theatre.

The combined geography is more striking. Corinth lies between Epidaurus and Delphi and only about 95 metres from the approximately 110.7 km Epidaurus–Delphi theatre-centre geodesic. The three theatre centres are therefore almost collinear at this geographical scale.

The architectural axis is close to the three-centre line but is not identical to it. This is a close geographical and directional correspondence, not an exact centre-to-centre sightline.

Compound-triplet test

Run 01C tested the Epidaurus–Corinth–Delphi case as a post-discovery compound configuration. It exhaustively ranked all eligible ordered triplets and then used 100,000 simulations per null model to ask how often an equal-or-better combination appears after orientation randomisation while retaining the observed geographic gate.

That observed-catalogue uniqueness does not translate into corrected statistical significance. In the exact-coordinate subset, two null models produced raw p-values below 0.05, but no Benjamini–Hochberg corrected q-value fell below 0.05. Full-corpus probabilities were approximately 0.146–0.165, and the most like-for-like eligible-source permutation gave approximately p = 0.165.

The configuration is therefore exceptional within the observed catalogue and worthy of investigation, but statistically suggestive rather than demonstrated.

Robustness and chronology

Coordinate-jitter tests from 25 to 250 metres preserve the basic geographical relationship, but none of the resulting statistical tests survives false-discovery correction. The published Epidaurus orientation uncertainty is approximately one arc-second and is negligible relative to the 0.67–0.81° directional residuals.

The principal historical weakness is chronology. The period fields used by Run 01C do not provide an overlapping construction interval for all three theatres. The present data therefore do not support a shared-building-phase claim, and the construction and reconstruction histories of Epidaurus, Corinth and Delphi require separate archaeological review.

Interpretation and limits

Theatre Orientation currently supports two distinct conclusions. First, the 58-theatre analysis does not demonstrate a general inter-theatre orientation network. Second, Epidaurus–Corinth–Delphi is the unique best observed compound triplet in the present catalogue and merits focused historical, topographic and archaeological investigation.

Neither conclusion establishes deliberate planning. Direction does not demonstrate intervisibility, near-collinearity does not establish common chronology, and a catalogue rank of first does not by itself establish rarity under every appropriate null model.

Next work and publication

• Verify the three theatre-centre coordinates against the strongest available monument plans or official geospatial records.
• Review construction and reconstruction phases of Epidaurus, Corinth and Delphi.
• Add topographic context or a terrain profile before making any claim about visibility.
• Check whether the Epidaurus axis has a nearer landscape or sanctuary target that could independently explain its orientation.
• Retain Run 01B as the negative general-network baseline and Run 01C as the completed compound-triplet test.

The statistical branch is complete; no Run 01D is planned with the present dataset. The appropriate publication route is a carefully qualified public paper, currently titled Where the Theatre of Epidaurus Looks, rather than a scientific paper claiming intentional design.

Website project summary v001 · ArchaeoLocus · 17 August 2026