//javjet

Navigation

Read the…

logo-mission-log155.gif
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 »

expand/close

General Flight Path Reconstruction

Re-navigation based on all available information

Research requirements demanded an extensive effort.
A fundamental research strategy was centered in an attempt to recreate the aerodynamic and environmental aircraft performance on the final flight, from well-established fact, well-founded inference, logical and experienced-based assumptions, and a critical application of statistical analysis of historical flight parameters, human factors and behaviors.

An area of inescapable uncertainty in the true location of this aircraft will always exist until a discovery is made. This research resulted in improvements in understanding the associated flight path, mission elapsed and endurance times, fuel consumption, and the probability for artifact detection.

Methodology
Our research methodology included a new approach to the navigation of the flight profile. Previous works generally used averages of total distance, divided by mission time, to ascertain location.

This research took a different approach by modeling the Electra with the following references

Flight performance data from the Lockheed 487 Report (L487)

Kelly Johnson telegrams of calculated and actual aircraft performance, and in-flight test data

Corroborated Lockheed 10A operating data, with virtually the same horsepower per pound of aircraft weight as the Lockheed 10E, slightly less frontal area due to smaller engine cowlings, and a Cambridge Fuel Analyzer of the type used by AE

Data from AE Electra flights prior to the World Flight

Historical statistical speed data from AE’s prior World Flight segments

Validated data from Long’s wind assessments

Swenson and Culick’s aircraft drag, speed and fuel consumption computational results

Aircraft and engine performance from operating manuals of aircraft using the same engine as in AE’s Electra (North American T-6, for example)

Fuel consumption analysis referencing actual Pratt-Whitney Specific Fuel Consumption (SFC) data for the R-1340-S3H1 engine, used by AE’s Lockheed 10E.

A software model was created in Jeppesen FliteStar flight planning software, and used to construct flight plans from Lae to Howland via three paths. Takeoff, climb, cruise, and descent were re-calculated by segments, with performance integrated manually from multiple sources. Fuel consumption was examined by profile segments, summed across the profile, and validated from multiple source data. A more precise, manual computational analysis of fuel consumption was made from resources, further refining this important factor.

expand/close

Overview - Flight Paths A, B and C

Assessment of three re-calculated flight paths

Research for this report includes assessment of three principal re-calculated paths, Path A, B, and C as defined below.

While each Flight Path will be examined in detail, in general, Flight Path A arrives almost everywhere, too early, and at 1912 GMT has actually over-flown Howland Island by enough to possibly preclude visual acquisition of the island, or the Itasca.

Flight Path B passes through the incorrectly reported 0519 GMT longitude position at 2 hours 18 minutes, and is then early at the 0718 GMT position report. AT 1912 GMT, this Path B arrives at the 1937 position coordinates for Howland Island.

Path B is also misaligned with navigation reporting position and time, and despite arriving at Howland Island, no person saw or heard the Electra. Path B may have passed through the 0519 GMT reported position, at an actual time of 0218 GMT, with these times misreported by Chater10 or Collopy11.

Reduced mission headwinds during the last 8 hours of the Lae to Howland segment, could result in Path B beyond Howland Island. A 5-10nm lateral error could result in nobody hearing or seeing the Electra, and the aircraft at a wind-adjusted Path B End-of-Navigation point, 20nm northeast-to-southeast of Howland Island.

The evidence suggests that the 0519 GMT reported longitude may be incorrect.

Correcting the 0519 GMT position report in longitude, with the actual position along the E 157.0 longitude, vice the E 150.7 longitude reported by Chater12, creates Path C. This path is reasonably aligned with all navigation reporting positions and times within 5% of the time the aircraft was at that point. Key factors such as entering the visual horizon to Nauru Island, where AE reported seeing lights on the island, are aligned in time and position. AE arrives slightly short of Howland Island due to what is possibly a navigational error, or miscalculation between 200nm and 100nm west of Howland Island, between 1745 GMT and 1815 GMT. One possible error is a sunrise celestial calculation of refraction, or dip angle computational error, of 31-70 nm depending on altitude at the time the fix was taken, such that “If the correction was not made, Noonan would have calculated that the Electra was nearer Howland Island than was the case.”13

Other errors are possible, including that FN made no errors and AE decided to descend slightly early, perhaps to keep Howland ahead of them to facilitate visual acquisition. This behavior would not be atypical for AE, as demonstrated on the Natal-Dakar segment when she turned opposite to FN’s suggested direction.

At the 1,000 feet altitude reported by AE approaching Howland Island, AE is at the edge of a visual acquisition range to the island and Itasca. Due to the rising sun directly ahead, visual acquisition would require being much closer to the Island.

Lateral track errors are possible, but there is no factual data upon which to make assessments of lateral navigation deviations from the planned course, and no evidence of lateral track error. On the contrary, the available data indicates AE adhered well to the track from Lae to Howland Island.

While all three paths are possible, Path A may be unlikely. Path B is possible, in that it deviates south of track for weather avoidance, and passes through the point chronicled at 0519 GMT, at an actual time of 0218 GMT. The numerals “5″ and “2″ could have been confused. Path C is likely.

The evidence for en route aircraft performance, mission times, position reporting, and key milestones, are all well aligned with navigation Path C. Even with reduced second-half mission winds, Path C concludes short of Howland Island, in the designed Primary Search Grid.

A final AE radio report at 2013 GMT with no further communication from AE, indicates a possible scenario in which the Electra contacted the water during terminal area maneuvering, perhaps due to pilot fatigue, loss of situational awareness, or due to fuel exhaustion, after 2013 GMT. A fuel consumption analysis (updated from Appendix 1) creates an endurance window until 2100 GMT. Fuel exhaustion between 2013 GMT and 2100 GMT is likely.

Evidence from Itasca weather reports for the morning of July 2, 1937 indicates light winds and a calm surface. Calm seas are difficult to fly over at lower altitudes because the pilot can lose awareness of altitude. At sea, the horizon and sea surface can blend into an uncertain mirage without sufficient detail to visually maintain desired altitude above the smooth water surface. Unintentional contact with the sea is a constant hazard during low altitude maneuvers over calm sea surfaces.

expand/close

End-of-Navigation Points

Where each flight path would have ended at 1912 GMT

An End-of-Navigation point (EON) was identified on each path, at 1912 GMT when AE thought, and reported, she had arrived at Howland Island (1937 coordinates). The End-of-Navigation point (EON) is determined by the lateral path, vertical profile dynamics, and aircraft performance. The EON point is the commencement point for terminal search maneuvering, and construction of search grids.

Wind effects and reasonable navigation errors were then considered with terminal maneuvering to create containment zones that comprise the Primary (west) and Secondary (east) Search Grid zones.
On the two most likely Paths, Path B and C, the effects of modified winds from 20 degrees left of the nose at 18 knots (approximately 25% less velocity) were examined to produce an error tolerance for the case in which AE held a magnetic course only, with no overnight wind correction applied.

A scenario examined the effect of a wind change for the last 8.5 hours of the mission.

A second scenario examined the effect of a wind change for the final 2.0 hours.

Reduced second-half winds are supported by data from weather forecasts from Hawaii, and surface vessel weather reports in the area of the flight. Both Hawaii preflight weather forecasts contained reduced second-half mission wind velocities.

Grid Search areas are containment zones accommodating these effects, which move the End-of-Navigation point slightly east, and slightly southeast, of the original track.

Milestone waypoints for AE position reports were placed on each path at the GMT times that AE made the report, to see where on the path, in time, these might have occurred. With consideration for tolerances in reporting behavior, variance in position reporting, and error in fixing positions, the aircraft locations over the earth at the times of the reports, support validation of the analysis. This helped provide context to path construction and timing.

EON Locations
All three paths are executed based on as much factual data as possible, concluding in End-of-Navigation points at time 1912 GMT. The location of the aircraft on each path, at this time, is shown below:

Path A is the great circle direct path from Lae to Howland Island, with an EON bearing from the island 066 degrees magnetic at 22nm past the 1937 Howland Island coordinates.

N 00° 54′ 22.2W176° 21′ 33.9

Path B passes through the waypoint reported by Chater at 0519 (a point with discrepancies in location and/or time), with an EON at the 1937 Howland Island coordinates.

N 00° 49′ 00.0W176° 43′ 00.0

Path C passes through a longitude-modified 0519 GMT waypoint with an EON bearing from the island 247 degrees magnetic at 35nm short of the 1937 Howland Island coordinates.

N 00° 40′ 51.7W177° 16′ 41.1

Path Depictions
The three paths are depicted below with the time of arrival at two important AE position reports. For the 0519 GMT and 0718 GMT waypoints, the aircraft could have arrived at the waypoint before the waypoint was reported, consistent with Fred’s navigation techniques demonstrated on the Oakland to Honolulu segment, and Amelia’s reporting of Fred’s waypoints on that flight.
While the Oakland-Hawaii segment revealed FN and AE waypoint arrival and reporting techniques, the Lae to Howland segment uniquely included passage over landmasses, unlike the Oakland-Honolulu and Natal-Dakar oceanic crossings. All three paths contain over-flight of good, visible island waypoints, where checks of position, time, and fuel consumption could have been made with good precision. On these unique segments, it is possible that a position report was issued shortly after establishing the aircraft at the waypoint, approximately 10-15 minutes later, a time that also was very close to AE’s pre-scheduled reporting at 15 and 45 past the hour.

Figure 1 Flight Paths.jpg

Figure 1 - Three Possible Paths and Initial Position Reports.

expand/close

Path C - Initial Discussion

Considered the most likely path flown

Path C is a likely path. Although it deviates south of other, more direct paths, this deviation is supported by two weather forecast reports. One report was delivered to AE in Lae on July 1, from Navy headquarters in Hawaii.14 AE received this report in hard copy. The second forecast arrived in Lae on July 2 during AE’s takeoff from Lae.15 This report was broadcast to AE on the hourly schedules arranged with Lae, and throughout a period of approximately 7 hours. Both pre-launch forecasts were for strong and dangerous thunderstorms east of Lae, on the direct path to Howland. The second report expanded the area of thunderstorms from 250 to 300 miles east of Lae, and provided an updated estimate of en route winds. While the first report contained wind estimates less than what AE reported at 0718 GMT, the second forecasted en route winds at “…east southeast about twenty five knots to Ontario, then east to east northeast about twenty knots to Howland….”

These winds estimates were surprisingly accurate, corroborated by AE’s 0718 GMT in-flight position report that included winds, at 23 knots, and from wind reports from Nauru Island, and surface vessels. Second-half mission winds were very likely at reduced velocity and from slightly left of the track from Lae to Howland.

These wind profiles were modeled by out team in the Jeppesen FliteStar software and in sensitivity analyses resulting in establishing the search grid.

AE was well aware of the existence of hazardous weather, between Lae and Howland. In fact, weather forecasts from Hawaii contained admonition to avoid flying through these dangerous thunderstorms. AE previously experienced heavy weather, from Natal to Dakar, and likely heeded Hawaii’s warnings.

Fred writes, in a letter to his friend, movie actor Eugene Pallette, “…The flight from Natal, Brazil to Africa produced the worst weather we have experienced - heavy rain and dense cloud formations necessitated flying blind for ten of the thirteen hours we were in flight.”16

AE also felt compelled to comment on the rain, “…the heaviest rain I ever saw. The heavens fairly opened. Tons of water descended, a buffeting weight bearing so heavily on the ship I could almost feel it.”17

Midday cumulous buildups over landmasses, such as the island of New Britain, may have also presented hazardous weather on the direct route to Howland Island that could be avoided with a relatively minor deviation southeast, across very good landmarks.

From their Atlantic crossing segment, Natal to Dakar, AE and FN possibly, and intentionally, planned a southerly deviation around New Guinea area weather, one with few penalties and several advantages.

Path C passes over Choiseul Island, the first island south of Bougainville Island. Both Bougainville and Choiseul are prominent visual landmarks. Bougainville’s mountains exceed 8,000 feet in the northern half of the island, but are easily avoided. Choiseul’s highest terrain is approximately 2,000 feet.

This deviation on Path C added only 42nm to the overall mission distance. The path also facilitated an afternoon setting-sun celestial fix, from the left side of the aircraft, inbound to the 0718 GMT reporting point near Nukumanu Island.

Figure 2.jpg

Figure 2 - Path depictions and supporting Factors for Path C.

Figure 3.jpg

Figure 3 - Pilot-eye view approaching the 0718 GMT position on Path C.

Only on Path C do all initial position reports in time coincide reasonably and closely with AE reported positions in space, and they agree in time within 5%. From the Oakland to Honolulu navigation logs, position reports were often made at some time after passing the reported position, and with the aircraft not co-located with the reported position. The lag between position passage, and reporting, is understandable in that it was AE’s first real navigation challenge working with Fred Noonan and Paul Mantz, and there were no landmarks corresponding to reported positions and times.

AE and FN may have sought to be more precise on the Lae to Howland segment, to more closely report positions and times. The data supports such an intention.

A more detailed analysis of Path C is contained in Part IV.

Back To Top

expand/close

Detailed Fuel Consumption Analysis

The Cambridge Fuel Analyzer’s key role

The detailed fuel consumption analysis, Appendix 1, initially results in up to 4 hours fuel remaining, at 1912 GMT. Further analysis since publication of Appendix 1 results in an upper boundary of 3 hours fuel remaining, at 1912 GMT.

Considering past AE behaviors, a lack of radio reports after 2013 GMT, and further analysis of an important instrument in the Electra (the Cambridge Fuel Analyzer), there is a high probability that only approximately 60 gallons of fuel remained at 1912 GMT - enough for 90 minutes flight time.

A fuel exhaustion time from 2013 GMT to 2100 GMT is highly likely.

This conclusion is well supported by the following analysis, completed after Appendix 1 was published, involving the Cambridge Fuel Analyzer.

A fundamental question has always plagued investigators and researchers, regarding theories of excessive fuel consumption on the Lae-Howland segment, arrival with far less than planned fuel reserves, and premature fuel exhaustion prior to landing on Howland Island.

The Cambridge Fuel Analyzer
The answer may very well be found in the Cambridge Fuel Analyzer, or CFA, instrument.

AE’s takeoff fuel quantity at Lae, according to Lockheed, Paul Mantz, Kelly Johnson, Swenson and Culick, and our own independent analysis, should have enabled the Electra to fly further and longer than it apparently flew - as much as 3 to 4 hours longer.

If AE was short of fuel, how would it be possible to burn more fuel than planned?

All prior researchers addressing this question concluded that excessive fuel consumption was due to one of the following

Incremental en route navigation adding distance to the planned route.

This would require adding hours of en route time to the original route distance.

Excessively high and inappropriate operating altitudes for the gross weight of the Electra, especially early in the mission, requiring excessive engine power and fuel consumption

Evidence from AE’s 0418 GMT position report at “height 7,000 feet,” the 0519 GMT position report at “height 10,000 feet,” and the 0718 GMT report at “height 8,000 feet” indicates that the aircraft is approximately at the optimum altitude prescribed by Kelly Johnson, and not high enough to produce excessive fuel consumption.

Excessive headwinds, well above forecasts, caused higher than planned power settings and resulting fuel consumption

Evidence exists to validate Long’s headwind value, which was initially more than forecast on 30 June and 1 July, but within the range of forecasted winds on 2 July.

2 July weather forecasts included winds that were expected to be reduced in the second-half of the mission, during the final 8 hours of the flight.

None of these traditional positions explain excessive fuel consumption, are all are considered not valid.

If AE arrived in the Howland area with critically low fuel quantity remaining, i.e., below planned reserves, there may be a more scientific, and more likely, explanation for excessive fuel consumption. This arises from understanding the importance of the Cambridge Fuel Analyzer equipment to achieving long range flight in the Lockheed Electra.

The Cambridge Fuel Analyzer (CFA), sometimes referred to as the Cambridge Exhaust Analyzer, was a very important tool in the World Flight attempt plan. The CFA monitors exhaust gases, enabling the pilot to precisely and safely set the optimum mixture control of the fuel-air mixture, to ratios that allow achieving optimum engine performance, minimum fuel consumption, and therefore, maximum aircraft range. The CFA assures that minimum fuel is consumed, with no adverse effect on engine health.

We showed in Appendix 1 that a fuel burn variance of just 1-2 GPH per engine, or 2-4 GPH total additional fuel consumption, could explain a significant fuel remaining variance.

Appendix 1 fuel remaining by our calculation was 123 gallons at 1912 GMT, enough fuel for 3 hours flight time. If an inoperative Cambridge Fuel Analyzer resulted in AE burning 4 GPH above plan (less than 10% variance) for 15 hours, the aircraft would arrive at 1912 GMT with fuel quantity at just 63 gallons, approximately 90 minutes fuel remaining.

An airline’s Electra Operating Manual states that without the CFA, the fuel-air mixture must be manually set to a more rich mixture, to prevent engine damage. The resulting higher fuel consumption reduces aircraft range. In fact, the Electra manual states that maximum chart ranges cannot be achieved without a functioning Cambridge Fuel Analyzer.

The CFA was used extensively on the Kelly Johnson test flights of AE’s aircraft, to maximize engine efficiency, and obtain the gallons per hour fuel consumption rate for different power settings. Every flight test data point contained an associated CFA value. These CFA, manifold pressure and propeller RPM settings were supplied to AE as mission profile specifications, including as detailed cruise specifications issued by Kelly Johnson for the World Flight.

The Lockheed 487 Report preamble contains the statements

“To enable close control to be maintained over the mixture strength, a Cambridge gas analyzer is connected into the exhaust system.”

“The complete performance has been computed conservatively based on actual flight test results on Model 10E. Fuel consumption data is based on results that have been obtained in flight with careful mixture control. To get a range of 4500 miles it will be necessary to calibrate the Cambridge Analyzer so that the fuel consumption curve shown on page 13 can be obtained.”

“The Cambridge Gas Analyzers should be carefully calibrated in flight to see if the fuel consumption data used in this analysis can be obtained.”

The report L487 was dated June 19, 1936, and subsequently, test flights were conducted by Kelly Johnson, using the calibrated Cambridge Gas Analyzer, to identify World Flight performance specifications for AE.

These recommendations were contained in three telegrams from Kelly Johnson to AE dated 11 March 1937.18

Clearly, the CFA was very important, and AE adhered to these CFA settings very closely for all flight profiles.

Figure 4.jpg

Figure 4 - Cambridge Fuel Analyzer

Apparently the CFA was also somewhat fragile, as it was frequently being repaired throughout the World Flight, at many of AE’s intermediate stops where maintenance was available. The leads to the exhaust stack, the analysis cells, and calibration were reported as problematic.

The CFA failed en route to Karachi, and on JUN 16, 1937 AE sent a telegram to George Putnam, “FUEL ANALYSER OUT ASCERTAIN FROM CAMBRIDGE INSTRUMENT IF POSSIBLE GET REPLACEMENT OR IF ANYONE AVAILABLE TO REPAIR HESITATE ATTEMPT PACIFIC WITHOUT [author's emphasis] CABLE CALCUTTA.”19

George Putnam replied, “KLM USES CAMBRIDGE CABLING AMSTERDAM HEADQUARTERS TO ARRANGE CALCUTTA SUPPLY NEW ANALYSIS CELL IF NECESSARY WHICH BELIEVE FAULTY STOP….”20

During 3 days of maintenance in Bandoeng, JUN 21, 22, and 23, the CFA was again, repaired. Among a long list of maintenance performed on AE’s Electra, specifically21

“Two broken leads in left analyzer cell of exhaust analyzer repaired.”

“Switch on junction box of exhaust analyzer repaired.”

“Transmitter on left engine of Eclipse flow meter repaired (soldering between pivot and internal magneto loose) and transmitter adjusted.”

. of Eclipse flow meter cleaned.

Oil and fuel filter strainers cleaned.

Thermocouple No. 3 lead, starboard engine repaired.

Thermocouple No.2 lead, port engine, replaced.

We know from AE’s logs that she then flew from Bandoeng to Surabaya, Indonesia, and the next day, flew back to Bandoeng for repairs to “an instrument necessary for long range flight.”22
In AE’s own words: “In the air, and afterward, we found that our mechanical troubles had not been cured. Certain further adjustments of faulty long-distance flying instruments were necessary, and so I had to do one of the most difficult things I had ever done in aviation. Instead of keeping on I turned back the next day to Bandoeng. With good weather ahead, the Electra herself working perfectly, and pilot and navigator eager to go, it was especially hard to have to be “sensible.” However, lack of essential instruments in working order would increase unduly the hazards ahead. At Bandoeng were the admirable Dutch technicians and equipment, and wisdom directed we should return for their friendly succor.”23

There is likely only one “instrument necessary for long range flight” that AE would not need on shorter range flights - the Cambridge Fuel Analyzer - and it may be the subject instrument referenced in AE’s single passage, and in letters written by Fred Noonan.

The Electra also contained Eclipse fuel flow meters, and while occasionally problematic throughout the World Flight, the fuel flow meters were not necessary, and were never more accurate than the CFA.
This single passage concerning an “instrument necessary for long range flight” was not further explained, or developed anywhere in the research, and no researchers have addressed this aspect of the mission.

This may represent a very important finding.

Fred also commented on this seemingly important, instrument. In letters he wrote to Ms. Helen Day, a friend in Miami, FN alluded to what was likely, the Cambridge Fuel Analyzer24

22 June 1937 - from Bandoeng, Java - “We arrived here yesterday from Singapore[,] and as some minor instrument adjustments were necessary we decided to remain here an additional day.”

27 June 1937 - FN writes again from Koepang, Timor Island, Dutch East Indies after arriving from Surabaya, Java, in which he references the previous few days, “…we spent considerably more time in Java than we expected to - had some minor but important instrument adjustments to be made, and as the Dutch Line is using the new DC3 Douglas - equipped with similar instruments - we decided to have the work done in their shops at Bandoeng. We remained there from last Sunday until yesterday - Saturday. Took off once and got as far as Surabaya - about three hundred and fifty miles - only to have the instruments fail again - so returned to Bandoeng. They are functioning perfectly now, thank goodness for the Dutch mechanics.”

In a short period of time, this “long range instrument,” which may be the Electra’s CFA, had recently failed inbound to Bandoeng. It likely failed again after leaving Bandoeng. And even following the return to Bandoeng, and repairs, the CFA failed just two flight segments later, from Darwin to Lae. It was serviced in Lae according to servicing records there, which detailed replacement of an analysis cell, which AE had aboard the Electra.25

The Lae Chief Engineer’s Report26 contains entries for repairs to AE’s Electra before embarking on the Lae to Howland Island segment.

Oil filters inspected and cleaned - both engines.

Fuel pump starboard engine removed and replaced.

Thermocouple connection on No. 4 cylinder, starboard engine, repaired.

New cartridge fitted to exhaust gas analyzer - starboard side

En route to Howland Island, it is quite possible that AE suffered yet another failure of the CFA, which would adversely affect fuel consumption.

Without the CFA, AE could not set optimum power, or minimum fuel consumption rates, nor attain maximum aircraft range for the fuel load from Lae to Howland.

At the point of failure, AE may have determined that returning to Lae, or perhaps, Bandoeng, for another repair of the CFA, was complicated by weather, and unnecessary if careful setting of the engine mixture was monitored. A return would require an excessive amount of time, perhaps deemed unacceptable to maintaining the arrival schedule in Hawaii of 4 July 1937.
For whatever reason, AE may have elected to continue without operable Cambridge Fuel Analyzers, perhaps on one, or both, engines. The result was likely increased fuel consumption, which resulted in arriving in the Howland area with perhaps half the quantity theoretically possible.

Throughout this research, attempts to acquire documentary evidence of the difference in fuel consumption between using and not using the Cambridge Fuel Analyzer, were unsuccessful. It is not clear that such data exists at all, or was ever compiled by the Cambridge Instrument Company27, Lockheed, or Pratt-Whitney.

Of note is that all of Kelly Johnson’s performance recommendations resulted from use of the CFA. It may have been so important that AE actually backtracked an entire flying day, and invested another ground maintenance day, to have this instrument “necessary for long range flight,” repaired.

Sperry Gyro Horizon
One other instrument that may be considered necessary for long-range flight could be the Sperry autopilot system. This equipment relieves the pilot from manually and continuously controlling the aircraft for many hours. Such relief is beneficial in fatigue management.

The Sperry was repaired in Lae.28

“Sperry Gyro Horizon (lateral and fore and aft level) removed, cleaned, oiled, and replaced, as this reported showing machine in right wing low position when actually horizontal.”

The Sperry autopilot equipment was problematic in Miami before commencing the flight to San Juan, Puerto Rico. Pan Am mechanics identified the problem, which was a faulty initial installation at Burbank, CA. They corrected the problems and the equipment worked perfectly leaving Miami for San Juan.29

But this equipment was never identified as problematic throughout the World Flight.

Separate redundant flight attitude instruments of which there were likely two, one for the left seat pilot and one for the right seat pilot, were necessary for instrument flight in clouds, and at night, during any flight segment, not specifically for “long range flight,” since the aircraft could be flown manually.

One artificial horizon instrument was repaired in Bandoeng30 (bar stuck in case).

It is possible that had the Sperry Gyro Horizon autopilot failed, depending at what time that occurred, AE may have been compelled to return to Lae, and abort the long distance, night, overwater segment to Howland. Since there were redundant and backup artificial horizon instruments to reference in manually flying the aircraft, it is doubtful a failure of the autopilot system alone would cause a turn back to Lae.

The most compelling evidence that the Cambridge Fuel Analyzer was the subject of the “instrument necessary for long range flight” was AE’s investment in having it repaired multiple times, and AE’s telegram to George Putnam, from Karachi, mentioned above.

As AE stated, “…HESITATE ATTEMPT PACIFIC WITHOUT [Cambridge Fuel Analyzer]…” and this largely substantiates that the CFA was indeed the necessary instrument, and that had it failed from Lae to Howland, would certainly have resulted in increased fuel consumption.

Appendix 1 Excerpt - Review and Summary
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 (”modified” 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 all available data, but largely from Kelley 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. AE should have arrived at time 1912 GMT with 123 gallons, enough for 3 hours 04 minutes endurance.

Fuel Remaining Implications
Our calculations of 957 gallons fuel consumed result in arriving in the Howland area with 123 gallons.

A failure of the Cambridge Fuel Analyzer would have increased fuel consumption, possibly accounting for the entire “over burn” of fuel, from the amount that should have produced a sufficiently comfortable range and endurance upon arrival to the Howland area, to a quantity that may have precipitated the initial check in at 1912 GMT “we should be on you, gas is running low.”

In this case, there would be no gross navigation errors, no large, unexpected headwinds, no extraordinary climbs to altitudes well in excess of that specified by Kelly Johnson, and no excessive cruise speeds maintained that would have irresponsibly burned an excessive amount of fuel. Such speeds would have been statistically abnormal, well outside prescribed ranges and inconsistent with AE’s past operating performance.

Under a Cambridge Fuel Analyzer failure scenario, the aircraft may have experienced fuel exhaustion between 2013 GMT and 2100 GMT.

Back To Top