Read the…

Experience the thrills and defeats of an authentic deep sea adventure as you immerse yourself in the CATALYST 2 Mission Log. More>>

Waitt Institute

The Waitt Institute is a non-profit research organization that serves as an exploration catalyst, enabling scientific pioneers to transform the ways in which discoveries are made. More »


Search Stragegy Refinement

Narrowing the search area

Search Strategy Refinement

AE - Amelia Earhart

BHP - Brake Horse Power

CFIT - Controlled Flight Into Terrain

EON - End of Navigation Point

FN - Fred Noonan

GPH - US Gallons Per Hour

LOP - Line of Position

L487 - Lockheed and Kelly Johnson Report 487

SFC - Specific Fuel Consumption (lbs/BHP/hr)

Search Grid and Scenarios
The challenge to identify a starting point in the overall search grid requires a detailed fuel consumption analysis and a consideration of lost aircraft location theories.
We agree with Long and other researchers that NR 16020 was not lost en route, did not land at Howland or Baker Island, and lacked sufficient fuel (shown later) to reach any other land mass. Therefore, we focus on a failure to arrive scenario, where the aircraft possibly crashed, experienced a Controlled Flight Into Terrain (CFIT) event, or exhausted its fuel supply.

A crash event could result from a loss of control, or a mechanical malfunction.

A CFIT event is an inadvertent collision with terrain (water), often involving a loss of situational awareness, but with the aircraft flying normally in terms of configuration, speed and attitude.

Flying over smooth water conditions reported by Itasca, is a challenge.

Depth perception is more difficult than while flying over rougher seas.

A fuel exhaustion event could produce a survivable, controlled water ditching.

Various calculations result in sufficient fuel at 1912 GMT to conduct a 1.5-4.0 hour search, but more likely a search in the range of 1.5-3.1 hours. After 2013 GMT, the total absence of radio communications is unusual, supporting two possibilities.

1.  The aircraft may have impacted the water prior to fuel exhaustion.

2.  Fuel exhaustion precluded further radio communications. This could result from a mission fuel over-burn for unknown reasons. For example

a. Zero fuel at 2030 GMT would indicate a mission over-burn of 71 gallons.

b. Zero fuel at 2100 GMT would indicate a mission over-burn of 51 gallons.

These examples represent a 2.4 to 3.5 GPH variance in total fuel consumption from planning calculations. The per-engine fuel use variance of 1.2 to 1.75 GPH is certainly possible.

A search strategy requires calculation of where the aircraft is in time and space, and how much fuel and time remained, after arriving in the Howland area at 1912 GMT.

Reference Grids
Reference points are plotted in the search grids, including the Path C End of Navigation (EON) point and a 2013 GMT position.

There is evidence that winds in the final 8.5 hours of the mission either decreased in headwind component, and/or shifted direction to come from slightly left of course and at reduced strength.

In order to address the effects of winds that may not have been detected or accounted for by AE and FN, an analysis was completed for a range of possible wind values. This analysis applied various realistic wind values to fixed headings the crew could have maintained. The resulting End of Navigation points are contained in the existing search grid.

The Search Plan is oriented along a 337-157 degrees magnetic compass heading, perpendicular to the planned magnetic ground track from Lae to Howland. The Search Plan accounts for possible cross-track error en route, as well as subsequent Line of Position (LOP) ground tracks in the terminal area as functions of true or magnetic tracks.

The LOP established by FN at Howland sunrise in preparation for AE’s “about 100 miles out” position report at 1815 GMT, was 337-157 degrees true.

We assess AE did not fly the LOP initially, nor until at least 1928 GMT (”…circling…”), when they reported flying the LOP at 2013 GMT.


This report is 61 minutes after initial arrival at where they thought Howland Island was, following an initial search of the area. After the “circling” report at 1928 GMT, they likely commenced flying the LOP tracks in a rectangular pattern, progressing further east on each LOP, while attempting to contact Itasca and visually acquire Howland Island.

Anxiety was reported in AE’s 2013 GMT radio transmission. It would be possible that under the circumstances, AE flew northwest on a heading of 337 degrees magnetic.
not accounting for a magnetic variation correction to FN’s LOP if it was indeed in degrees True.

The 10-degree difference between true and magnetic courses in search grid orientation was constructed to examine positional effects on search operations.

The effect of this orientation is insignificant.

Back To Top


Fuel Remaining

The critical factor reviewed in depth

The amount of fuel remaining in the Electra at 1912 GMT is important because it determines how long the aircraft could stay airborne, and how far it could fly, before fuel exhaustion.

The amount of fuel remaining is a function of how much fuel was consumed.

This analysis is a challenge due to the need for estimation in the absence of empirical data.

Direct evidence from AE, and World Flight data, corroborate that mission segments were flown adhering closely to the accurate Kelly Johnson specifications.

It was not possible to conduct test flights on the aircraft to acquire necessary performance data because there are very few remaining Lockheed Electra 10E aircraft. The one article we photo-documented is not flyable.

In order to make an assessment of fuel consumption, and the possible amount remaining, other data were examined.

Fuel Consumption
Constants used in this analysis are

Fuel weight is 6 pounds per gallon (also used by Swenson and Culick).

Maximum endurance speed is 120 mph at 40 GPH.

Estimated from L487 and Pratt-Whitney engine data (35-40 GPH).

Takeoff fuel quantity is 1080 US gallons (Swenson and Culick).

The following resources were examined

Kelly Johnson Telegrams.

“Range Study of Lockheed Electra Bi-Motor Airplane,” Lockheed Report 487 dated 4 June 1936.

AE Flight Notes, Flight Performance history, and direct evidence of performance.

“Aircraft Engine Characteristics Summary,” Pratt-Whitney dated 1 May 1951.

“Flight Operation Instruction Chart,” detailing aircraft performance for the North American AT-6 SNJ with the Pratt-Whitney R-1340-AN-1 (very similar to -S3H1).

Jeppesen Flight Planning software interpolating L487, Kelly Johnson and AE Flight Notes data.

Each of these resources is limited to some extent, in its usefulness.

Lockheed Report 487 is largely analytical, reflecting computations vice actual aircraft performance. The L487 Report includes brake horsepower, however, it does not specifically cover engine power settings frequently used and reported by AE in her flight notes.

None of the three Kelly Johnson telegrams (issued post-L487 Report) containing flight test data mention the gross weight of the Electra as tested, the speed associated with each power setting, outside air temperature, or the brake horsepower at which fuel consumption data were recorded.

From the Kelly Johnson flight test data, notably the third Telegram to AE, mission profile recommendations were made for altitude, power setting, and fuel consumption. These recommended settings were grouped into 3-hour segments, reflecting that Electra aircraft performance is relatively unaffected by small gross weight changes.

Aircraft Gross Weight
The gross weight of the aircraft and the power from its engines are important to climb rate, cruise speed, and fuel consumption.

For constant altitude, constant speed cruise flight, lift must balance weight, and engine power must balance the total vehicle drag from all sources.

An aircraft design axiom is that an “aircraft climbs on its engines.” Excess engine power beyond that required for cruise flight where vehicle drag is balanced by engine power, can be used to climb, and where excess power is limited, climb rate is also limited.

Report L487 indicates climb rates at AE’s operating weight, should be in the range of 600-700 feet per minute. However, Pellegreno reported routinely requiring 20 minutes to reach 1,000 feet, after takeoff. L487 optimum initial cruise altitudes were 2,000-4,000 feet. Kelly Johnson modified the initial cruise altitude to 8,000 feet.

In practice, climbing the heavy Electra to 8,000 feet, 10,000 feet, or higher, likely required 30-60 minutes at high power settings of approximately 500-550 BHP and high fuel consumption ranging from 95-110 GPH (Pratt-Whitney).

AE’s Electra gross weight is important because the mission routinely operated at high gross weight. Takeoff from Lae was at approximately 15,500 pounds, 47.6% above design maximum gross weight. The Lae to Howland mission average gross weight was approximately 22.9% above design maximum gross weight.

One concern was that if the Kelly Johnson flight tests were performed at significantly “lower-than-actual-mission” aircraft weights, the resulting profile recommendations could result in AE experiencing less climb rate, slower aircraft speed, and higher fuel consumption throughout the World Flight.

Our assessment is that Kelly Johnson’s data was accurate.

The empirical data from Kelly Johnson’s flight tests are very close to Pratt-Whitney engine data, and likely resulted from either testing the Electra at actual operational weights, or from computational corrections to test data, producing the three Telegrams recommending the following fuel consumption planning data.

3 hours at 60 GPH for 180 gallons

3 hours at 51 GPH for 153 gallons

3 hours at 43 GPH for 129 gallons

10 hours at 38 GPH for 380 gallons

This data agrees well with fuel consumption data we derived by calculations from Pratt-Whitney engine data supplied by the Smithsonian Institution, and from flight handbooks for other aircraft using the same engine as in AE’s Electra.

Kelly Johnson specified settings for 3-hour cruise segments, however, he did not specify takeoff, climb or descent fuel consumption data.

Combining Pratt-Whitney engine data, with Kelly Johnson’s recommendations and data, offers a more complete profile of fuel consumption.

Takeoff using 10 gallons

1 hour climb using 110 gallons

3 hours at 60 GPH using 180 gallons

3 hours at 51 GPH using 153 gallons

3 hours at 43 GPH using 129 gallons

8.5 hours at 38 GPH using 323 gallons

0.5 hours descent at 30 GPH (estimated) using 15 gallons

Total 920 gallons required from takeoff to 1912 GMT

Fuel Consumption and Time Remaining From All Analyses
Below summarizes the solutions for mission fuel consumed, and fuel remaining upon arrival at where AE thought Howland should be, at 1912 GMT.

If AE began with 1080 gallons, and flew this Kelly Johnson profile (”adjusted” for takeoff, climb and descent) requiring 920 gallons, the total fuel remaining would have been 160 gallons at 1912 GMT, or 4 hours endurance.

Computer flight profile modeling of data largely from Kelly Johnson and L487 data, also indicate the total fuel remaining would have been 160 gallons at 1912 GMT, or 4 hours endurance.

While our Jeppesen software model results, in terms of fuel used, corroborate the Kelly Johnson/L487 Report, our further analysis offers increased accuracy in this area.

Swenson and Culick’s analysis concluded that AE had enough fuel for 20 hours 38 minutes total mission time. Subtracting the known mission time of 19 hours 12 minutes, results in approximately 1 hour 26 minutes remaining endurance.

This represents 57.3 gallons remaining at 1912 GMT.

Our research, using a specific flight profile segment analysis technique, results in a total mission fuel burn of 957 gallons. Under the best circumstances AE should have arrived at time 1912 GMT with 123 gallons, enough for 3 hours 04 minutes endurance.

This figure could have been reduced due to malfunction of the Cambridge Fuel Analyzer and increased fuel consumption en route due to cruise altitude choices, winds or other factors.

“…Gas is Running Low…”
For the Oakland to Honolulu flight, AE planned to have 4 hours of fuel remaining, a “…good safety margin…” by her own account. AE also planned on having enough fuel on the Honolulu to Howland flight, 8 hours fuel remaining, to return to Honolulu from Howland in the event of an aborted landing situation.
Below 4 hours fuel quantity, even were it planned, AE would likely consider this a low fuel quantity situation, as she reported at 1912 GMT.

With fuel remaining of 4 hours, per Kelly Johnson/L487, and software modeling solutions, it is not likely that AE would have reported a low fuel condition.

With fuel remaining of 1 hour 26 minutes, per Swenson and Culick, it is likely AE would have reported a low fuel condition, but possibly with a sense of urgency, due to the extreme nature of her fuel quantity, with just 57.6 total gallons remaining.

Our analysis of 3 hours 04 minutes fuel remaining falls within the range for AE to report a low fuel condition, but without a sense of urgency. This figure could have been reduced due to malfunction of the Cambridge Fuel Analyzer and increased fuel consumption en route due to cruise altitude choices, winds or other factors.

Back To Top


Fuel Consumption

Conclusion of the fuel consumption analysis

Kelly Johnson’s third Telegram modifying World Flight operating parameters is important because indications from direct evidence are that AE closely adhered to these recommendations for altitude, and for mid-segment power setting, speed and fuel consumption, throughout the World Flight.

Unfortunately, AE made no reference to actual fuel burn, and drawing conclusions in this area requires analysis from a broad spectrum of data.

Assembling various aircraft performance data elements from pre-mission preparations, and 30 days of the World Flight, provides information on engine settings, speeds, altitudes, etc. These can be used to assess BHP, from which specific fuel consumption (SFC) can be derived.

The engine’s specific fuel consumption (SFC) is an engineering parameter, defined as “pounds of fuel per brake horsepower per hour.” Since this term is difficult to put into perspective, a more useful metric is gallons per hour (GPH).

SFC can be directly converted to GPH, and related to miles per gallon, range, and endurance.

We examined Pratt-Whitney documents from the Smithsonian Institute for AE’s engines, Swenson and Culick’s SFC calculations, and flight handbook engine operating data for nearly the same engine installed in a North American T-6 single engine aircraft, as well as for the Lockheed 10A aircraft, in context with all other performance data.

While the lack of Electra 10E-specific operating data hampered the investigation, we found that published operating information also varies among sources. Examples include

For takeoff, the Pratt-Whitney R-1340-S3H1 engine has a 5-minute time limit at that power setting, per Pratt-Whitney documents.

The L487 report specifies setting takeoff power for 1 minute, then directs a power reduction “…as soon as it is safe…” (L487 p6). It does not define climb conditions.

Swenson and Culick discusses a climb power setting of 420 brake horsepower (BHP). The source of this specification is not provided.

Pratt-Whitney specifies 550 BHP for climb, in engine data charts.

For cruise power, Swenson and Culick does not discuss power settings. L487 specifies for AE’s initial gross weight, a cruise power setting must be 375-400 BHP.
The graph below is plotted from Pratt-Whitney engine data for AE’s engines. The GPH curve (parabola) is relatively flat in the range of 250-400 BHP typically used for cruise flight, producing a linear relationship of BHP and GPH.

Figure 16.jpg

Figure 16 - From Pratt-Whitney Engine Data

BHP setting is important to a mission fuel analysis, because fuel consumption is directly proportional to BHP and gross weight.

At initial heavy gross weights, higher BHP in the range of 375-400 BHP is required, while at lower gross weights (achieved at approximately one third of the mission distance) power can be reduced to more economical settings, such as 250 BHP, to maintain prescribed speeds and altitudes, and achieved desired ranges.

The table below compares sources of fuel consumption data, with the associated effect on the mission fuel used (Total Gallons Required).

Table 6 - Engine Fuel Consumption

As corroboration, the P&W R-1340-AN-1 engine in the North American T-6 aircraft achieves its best long-range cruise at 5,000-10,000 feet altitude, burning 22-23 GPH. For simplicity, we can double this value, to approximate a representative cruise value for the twin-engine Electra of 44-46 GPH.
Differences among source data are relatively reasonable for the 1937 period

AT-6 engine GPH data, doubled to approximate Electra fuel consumption, is within 10% of simple (non-weighted) averages for Kelly Johnson and Pratt-Whitney data.

Kelly Johnson and Pratt-Whitney GPH data are within 3% to 7%.

L487 specifies initial cruise fuel consumption of 57 GPH

Within 4%-8% of Pratt-Whitney data.

Within 5%-11% of Kelly Johnson data.

Kelly Johnson and L487 cruise fuel burn at 250 BHP is 39.2 GPH

Within 4% of Pratt-Whitney’s published chart data.

Two-thirds of the Lae to Howland mission was specified to be flown at 250 BHP, which was also used by AE during the Natal to Dakar Atlantic Ocean crossing, earlier on the World Flight.

Pratt-Whitney engine data examined, to date, is for standard conditions of pressure and temperature. The Lae to Howland environmental conditions in temperature were warmer than sea level standard, which increases fuel flow. Flying at high altitude, while not exceeding optimum altitude, has a small positive effect on reducing fuel flow for AE’s engines.
These effects could account for differences between Pratt-Whitney engine data, and Kelly Johnson flight test data.

MSI Analysis (Multi-Source Integration)
Despite the absence of specific Lockheed 10E Flight Manuals, flight tests, and variances in published data, we can reach reasonable conclusions using all available resources.

After departing Lae with 1080 US gallons, and arriving in the Howland area at 1912 GMT, Fuel Remaining values from Kelly Johnson, L487, and our specific analyses, are all among the 9 unique results from Swenson and Culick’s interesting and comprehensive sensitivity analysis.

These 9 unique results are shown on the following graph. These serve as increasing confidence in identifying reasonable values for fuel consumption, fuel remaining, endurance time after 1912 GMT, and where the Electra could be located.

Figure 17.jpg

Figure 17 - Swenson and Culick conclusions (Average 21.37 hours. Standard Deviation 1.58 hours)

Conclusions for Fuel Consumption
The conclusions from this analysis include

AE had sufficient fuel for the Lae to Howland flight under existing environmental conditions, for Paths A, B and C, plus adequate reserves for a terminal area search.

After 2013 GMT there were no further transmissions heard from NR 16020.

Given AE’s communication history, this is uncharacteristic.

AE transmissions are expected at 2030 GMT, 2045 GMT, and 2100 GMT.

This supports a theory of a pre-fuel exhaustion water impact, possibly between 2013 GMT and 2100 GMT.

This may result from a CFIT event, or a mission fuel over-burn.

The fuel consumption rates in GPH computed from Kelly Johnson, L487, and Pratt-Whitney engine data, reasonably agree with empirical data from AE flight logs and position reports, in the range of 250 BHP prescribed for mission cruise.

A summation of discrete, mission segment analyses can produce a more accurate result in fuel consumption, using BHP and SFC with all other data.

Reaching Gardner Island, at approximately 400 statute miles distant, at 120 mph would require 3 hours 22 minutes, after 2013 GMT, or 4 hours 22 minutes after arriving at Howland Island at 1912 GMT.

Only Case 4 and 5 in Swenson and Culick’s analysis enable this result.

Back To Top


Possible Impact Area

Additional conclusions to determine search grid

At 1912 GMT the three basic fuel remaining scenarios are

1 hour 26 minutes

3 hours 04 minutes (and reductions due to possible failure of the Cambridge Fuel Analyzer)

4 hours 00 minutes

Position estimates result from search maneuvering and estimates of aircraft position in time.

Key points include

At the Path C EON point, the aircraft flys west for 10sm, then east for 10sm, then east another 10sm. At that point, AE reported, “circling,” and embarks on flying the LOP as a magnetic compass heading 337 degrees.

The LOP is flown for 20 minutes, covering 40sm.

A turn east then to a compass heading of 157 degrees, requires 6-7 minutes, where the aircraft then searches southeast.

This pattern is continued until reaching a 4-hour fuel exhaustion point.

Three key locations are added to the search area, corresponding to fuel remaining calculations, from Swenson and Culick, Kelly Johnson/L487, and our analysis.
Key inferences from this analysis include

It is realistic to expect further AE radio reports from 2030 GMT to 2100 GMT, or later.

The aircraft would be located in the search grid at 2100 GMT.

Of the three fuel remaining calculation scenarios

The Kelly Johnson/L487 point is considered least likely.

The Swenson and Culick point is considered possible.

A point between our most optimistic calculation and the Swenson and Culick result is considered the most likely of the fuel exhaustion scenarios, allowing the possibility of an en route failure of the Cambridge Fuel Analyzer, slightly increased fuel consumption rates, and reduced fuel remaining in the Howland Island area.

Search Considerations
If AE used fuel differently from this analysis, she likely used more fuel, not less fuel, resulting in fuel exhaustion in less than 3 hours 04 minutes, and within the Search Grid.
Effects of Significant Lateral Deviation North of Path C

If the aircraft passed overhead Nauru Island, it would mean that navigation had either inadvertently deviated 120nm in just three hours since flying over Nukumanu Island, or the course deviation was intentional to facilitate sighting Nauru’s expected lights, and logging the last good navigation fix until Howland Island.

An inadvertent deviation of this magnitude is very unlikely.

Intentionally passing overhead the island to establish a position is a reasonable intention. Unfortunately, there is no evidence to support the theory.

If we suppose that it did occur, then the course to Howland Island from overhead Nauru Island converges with the Path C track, terminating within a few miles northwest of the Path C End-of-Navigation point.

No matter how close to Nauru Island the Electra passed, at 1912 GMT on Path C it would be located within the Search Grid, very near the Path C EON point.

From a position overhead Nauru Island, another possible course would parallel Path C to 1912 GMT.

This is considered unlikely, as it would indicate intentional navigation to a point other than Howland Island, or a failure to correctly navigate to Howland Island.

The effects of reasonable lateral deviations place the aircraft in the existing Search Grid for scenarios of wind and weather considered possible, or likely.

Back To Top