NEAPOLIS FORGOTTEN PATHS RESEARCH INSTITUTE | PRE-EXPEDITION WORKING PAPER
This working paper was completed and publicly deposited prior to the planned October 2026 field investigation of the Taraia Object at Nikumaroro.
THE TWO-ISLAND DECISION HYPOTHESIS – Version V 1.0
A Pre-Expedition Behavioral, Navigational, and Physical Constraint Model for Amelia Earhart’s Possible Phoenix Islands Landfall
Author: Douglas Estill
The Neapolis Forgotten Paths Research Institute
**This working paper was completed and publicly deposited prior to the planned October 2026 field investigation of the Taraia Object at Nikumaroro.**
Companion website link – Zenodo – DOI: 222566702
**This paper freezes the model, assumptions, predictions, and falsifiers before the planned October 2026 field investigation. It is not a claim that Earhart reached Gardner/Nikumaroro or that the Taraia Object is NR16020.**
Abstract
This working paper develops a two-island decision model for the final phase of Amelia Earhart and Fred Noonan’s 2 July 1937 flight. Rather than beginning with a preferred island, the model treats navigation, fuel, visibility, pilot behavior, landing physics, post-landing survival, and radio survivability as interacting constraints. Under the specific working branch tested here, an underestimated headwind leaves the Electra roughly 130 nautical miles west of Howland near 19:12 GCT. Continued flight southeast on the reported 157°/337° line produces an unusual convergence: near 20:55 GCT, McKean and Gardner are nearly equidistant at approximately 192 nautical miles, while a later Sun observation becomes more useful as a crossing line. A climb to approximately 5,000 feet materially increases the visual horizon without consuming prohibitive fuel in a late-flight, much lighter aircraft. The model then asks a behavioral question: if both islands can be evaluated from altitude, would experienced aviators optimize for the safest touchdown or for the best overall chance of survival afterward? McKean appears to offer the safer aircraft landing; Gardner appears to offer greater perceived post-landing resources but a more hazardous reef or shallow-water landing. The paper therefore registers two falsifiable Gardner branches: a direct Gardner commitment and a McKean-reconnaissance-first branch followed by a deliberate Gardner diversion. If future investigation identifies NR16020 inside the Gardner lagoon, this model predicts evidence of an earlier reef/beach grounding and subsequent transport rather than an original deep-water lagoon ditching.
1. Evidence Discipline and Scope
The model is deliberately conditional. The numerical results below are not presented as historical facts; they are consequences of explicit assumptions. Each assumption can be changed and the model rerun.
The evidentiary order used here is: (1) contemporaneous government or service records; (2) contemporary technical documentation; (3) later technical reconstruction; (4) modern hypothesis literature. The model does not treat any modern Gardner/Nikumaroro theory as established fact.
The purpose is not to make Gardner fit. The purpose is to determine whether a coherent, physically possible and behaviorally intelligible sequence survives attempts to falsify it, and to identify observations that would disprove it.
2. Working Assumptions Used in This Version
| Variable | Working value | Status |
| 19:12 GCT position | ~130 NM west of Howland on a parallel 157°/337° LOP | Model-derived from the headwind/ground speed branch |
| Late-flight airspeed | ~130 kt | Working economy/search value |
| Late-flight fuel burn | ~38 gph | Long-endurance Johnson-style branch |
| Fuel at 19:12 | ~188 gal | Model-derived |
| 19:12–20:13 action | Continue generally southeast on 157° | Behavioral/navigation branch, not proven |
| Search altitude change | Climb from 1,000 ft to ~5,000 ft | Behavioral hypothesis |
| Late-flight climb example | ~1,000 fpm; ~110 kt horizontal; ~100 gph during climb | Sensitivity assumption for a much lighter aircraft |
| Second-Sun decision window | ~20:50–20:58 GCT | Model-derived geometry |
3. The Geometric Convergence: McKean and Gardner Become Nearly Equidistant
The most consequential result of the geometry is not that Gardner is “on” the 157° line. It is that, under the working western-shortfall branch, continued southeast movement changes the relative cost of the two nearest serious Phoenix candidates. At 20:13 GCT McKean is approximately 253 NM away and Gardner approximately 270 NM away. By about 20:53 the distances are ~194.4 and ~195.2 NM. The equal-distance crossover occurs at approximately 20:54–20:55, at roughly 192 NM to either island. After that point Gardner becomes marginally closer.
That crossover falls inside the same interval in which a later Sun line has rotated enough to become materially more useful as a crossing line. The coincidence does not prove a second sight was taken; it does make the decision problem unusually well constrained in time.
Figure 1. Under the working 157° southeast branch, McKean’s initial distance advantage disappears near 20:55 GCT.
4. Why 5,000 Feet Matters
At 1,000 feet the geometric horizon is only about 34 NM. At 5,000 feet it is roughly 75 NM. For a crew uncertain of position and seeing no land, a climb can be interpreted as a rational search action: it increases the visual search footprint and improves the plan-view presentation of low atolls.
Because NR16020 was thousands of pounds lighter late in the flight than at Lae takeoff, the 200–250 fpm departure-phase climb rate is not an appropriate fixed assumption for this later segment. A representative sensitivity case of 1,000 fpm gains 4,000 feet in about four minutes. At a conservative 100 gph climb setting, that costs approximately 6.7 gallons. This is a model assumption rather than a published late-flight NR16020 climb test.
Figure 2. Representative 5,000-ft branch after a ~20:53 turn toward Gardner. Gardner enters the geometric horizon first; McKean becomes simultaneously available later and farther off-axis.
5. The Two-Island Decision as a Human Problem
McKean and Gardner solve different survival problems. McKean is small, flat and barren. Gardner is larger, vegetated, lagoon-bearing, and visually marked by the wreck of the Norwich City. The Navy’s 1937 search descriptions capture this contrast clearly. Lambrecht described McKean as perfectly flat, about one square mile, without vegetation; Friedell judged that an airplane could have made a safe crash landing on the beach or in the center. Gardner was described as a large atoll, tropical vegetation covering the island, coconut groves at the western end, and the conspicuous wreck of a roughly 4,000-ton steamer on the reef.
The decision criterion therefore changes depending on what the crew is optimizing. McKean appears to maximize the probability of surviving touchdown. Gardner may appear to maximize the probability of surviving after touchdown: shade, vegetation, a lagoon, salvage from a wreck, and the visual suggestion—possibly misleading—of fresh-water or other resources. This distinction is central to the behavioral model.
| Observed from altitude | McKean | Gardner |
| Landing surface impression | Flat, barren, conventional forced-landing candidate | Reef/lagoon/vegetation; landing surface more complex |
| Shelter/resources impression | Poor | Much better apparent resources and shade |
| 1937 Navy assessment | Safe crash landing judged possible | Forced landing judged possible, but terrain more complex |
| 1938 aviation survey | N/A in this paper | Gardner ultimately judged a poor aviation site; tidal flats, coral irregularities, vegetation, reef hazards |
| Primary behavioral attraction | Protect airplane/occupants during touchdown | Increase perceived post-landing survivability |
6. Two Falsifiable Gardner Branches
Branch A: Direct Gardner Commitment
If the crew commits to Gardner near the second-sight/equidistance window, the model produces an arrival near the end of the favorable reef-landing window used in modern tide reconstructions. Under the long-endurance branch, substantial fuel remains. The strength of this branch is timing; its weakness is that it requires a specific navigational decision without direct documentary evidence of a second sight.
Branch B: McKean Reconnaissance First, Then Gardner
A more behaviorally explicit alternative is that the crew first inspects the visually safer flat island. A single circle would likely be enough for experienced aviators to decide that McKean is landable but resource-poor. They then inspect or commit to Gardner because its vegetation, wreck, lagoon, and apparent resources increase expected survival after landing. This branch consumes more time and shifts Gardner arrival later, likely beyond the most favorable dry-reef tide window. It therefore predicts a more difficult shallow-water or flooded-reef landing rather than the cleanest dry-reef scenario.
7. Landing Mechanics: Aircraft Survival Versus Radio Survival
The Model 10 landing gear was designed so that a portion of the main tires remained exposed when retracted, providing some protection in a forced gear-up landing. This makes an intentional gear-up McKean landing more mechanically plausible than a simple “belly plow” model. The remaining uncertainty is the actual 1937 surface: hard coral, loose coral, holes, ridges, and asymmetric snags could still produce severe propeller, nacelle, or structural damage.
Gardner creates the opposite tradeoff. A shallow-water or flooded-reef landing can dissipate energy, but water depth and reef irregularity can produce strong asymmetric drag. If post-loss radio transmissions are genuine, the aircraft cannot simply finish in deep water. The low-mounted electrical system imposes a physical condition: the aircraft must come to rest high enough, at least temporarily, for critical electrical components to remain dry.
Figure 3. Illustrative McKean forced-landing profile: powered descent from 5,000 to 1,000 ft followed by a short gear-up dead-stick final. Glide ratios are stress-test assumptions, not a published Lockheed 10E glide specification.
8. What the Taraia Object Can Test
The Archaeological Legacy Institute describes the Taraia Object as a shallow-water anomaly on the lagoon side of Nikumaroro, directly east of Tatiman Passage. Its current field plan targets an October 7, 2026 departure to investigate the anomaly using archaeological and remote-sensing methods. The object is not presently identified as an aircraft; its nature remains unknown.
This model makes a specific pre-expedition prediction if the Taraia Object proves to be NR16020: the aircraft probably did not make its original landing at the Taraia location. A post-loss-radio-compatible Gardner scenario requires an initial reef/beach position sufficiently elevated to keep critical electrical equipment usable. Therefore a confirmed Electra in the lagoon would imply a secondary transport history—reef grounding first, movement later—rather than an original deep-water lagoon ditching.
NEAPOLIS FORGOTTEN PATHS RESEARCH INSTITUTE PRE-EXPEDITION WORKING PAPER
9. Pre-Expedition Predictions and Falsifiers
| Prediction registered before fieldwork | Observation that would support it | Observation that would challenge/falsify it |
| If Taraia is NR16020, the lagoon location is secondary, not the original touchdown point. | Damage/deposition consistent with reef abrasion, breakup, or transport before burial. | Evidence of an intact, primary lagoon ditching with no plausible prior reef/beach phase. |
| A Gardner radio-compatible landing requires temporary elevation above damaging water depth. | Electrical/radio components show a period of survival before flooding/corrosion; wreck history indicates delayed inundation. | Immediate deep-water immersion incompatible with the professional post-loss radio chronology. |
| McKean is the more attractive aircraft-landing option; Gardner is the more attractive perceived human-survival option. | Independent evidence of deliberate island comparison, reconnaissance, or a later Gardner commitment. | Evidence fixing the crew on a direct one-island course with no practical opportunity to compare alternatives. |
| The 20:50–20:58 interval is a key navigation-decision window in this model. | New navigation evidence consistent with continued 157° southeast motion and a later corrective decision. | Primary evidence fixing a materially different position, heading, or fuel state during this interval. |
| A confirmed Gardner aircraft should not erase the McKean branch from the reconstruction. | Damage/flight evidence consistent with reconnaissance or a late diversion between the islands. | Evidence demonstrating no fuel/time margin for any McKean inspection or diversion. |
10. Interpretation
The central contribution of the Two-Island Decision Hypothesis is methodological. It replaces the question “Which island did Earhart choose?” with a sequence of constrained decisions: what could the crew know, what could they see, what could the airplane still do, and what survival tradeoff would two exhausted aviators plausibly make? The result is not a proof of Gardner. It is a structured decision tree that can be attacked by new evidence.
The model also clarifies why McKean and Gardner should not be treated as mutually exclusive navigation hypotheses. A crew could visually evaluate one and ultimately land at the other. Likewise, a future discovery of an Electra component at Taraia would not, by itself, establish the initial landing point. The physical history between touchdown and final deposition would still have to be reconstructed.
Conclusion
This paper is intentionally dated before the October 2026 Taraia expedition. Its purpose is to prevent post-discovery reasoning from being mistaken for prediction. The key registered propositions are: (1) a late 157° southeast branch can make McKean and Gardner essentially equidistant near 20:55 GCT; (2) a climb to approximately 5,000 feet is a behaviorally rational visual-search action and is not prohibitive under a lighter-aircraft fuel model; (3) McKean and Gardner present different survival affordances—safer-looking touchdown versus richer-looking post-landing environment; (4) an experienced crew could reasonably inspect one island and commit to the other; and (5) if a Gardner/Nikumaroro wreck is ultimately found inside the lagoon, a radio-compatible scenario predicts an earlier reef/beach grounding and later transport.
The forthcoming fieldwork can therefore do more than identify an object. It can test whether the physical history of that object is compatible with a pre-registered navigation, behavior, landing, radio-survival, and transport model.
Selected References
4. New Zealand Pacific Aviation Survey Expedition, General Report, 1939. Transcription hosted by TIGHAR.
5. Lockheed Aircraft Corporation, Report 487, “Range Study of Lockheed Electra Bimotor Airplane,” C. L. Johnson, 19 June 1936.
6. “The Electra,” contemporary Model 10 description reprinted by Aviadejavu; notes that a small portion of the tire remained exposed with the gear retracted for forced-landing protection.
7. Archaeological Legacy Institute, “The Taraia Object: Amelia Earhart’s Aircraft?” current project page, accessed 6 September 2026. The project lists a target fieldwork departure date of 7 October 2026 and describes the object as an unconfirmed lagoon anomaly east of Tatiman Passage.
8. Purdue University / Purdue Research Foundation public materials on the 2026 Taraia Object expedition and Amelia Earhart research initiative.
Model note: All distances, fuel values, arrival times, sighting windows, and glide examples identified as model-derived are analytical reconstructions used to test internal consistency. They should not be cited as primary historical facts.
RELATED ARTICLES: