Serious consideration was given to the possibility that the significant changes in aircraft performance were the result of a pre-impact failure or malfunction of the aircraft. Extensive and detailed examinations were conducted on all the recovered wreckage. Although much of the aircraft was consumed in the post-crash fire and the complete integrity of most of the aircraft systems could not be determined, the Board was unable to identify any physical evidence of such a failure or malfunction. All damage to the aircraft and its components was assessed to be the result of impact and the post-crash fire. The aircraft configuration at impact was determined to be normal for the planned take-off.

There was, however, considerable information in the form of witness statements which suggested that problems with the aircraft were present before the accident. Specifically, these related to the flight control system, the hydraulic system, the number four engine, and the thrust reversers. In addition, there were the reports of the yellow/orange glow emanating from the underside of the aircraft and the evidence of a lower rpm of the number four engine at impact. In the absence of FDR information pertaining to aircraft system operation and because of the extensive destruction of the aircraft which precluded a complete examination of all aircraft components, several possible malfunctions were analysed to determine their likelihood and what impact, if any, they would have had upon the accident flight.

The reported binding and ratchetting of the co-pilot's control column suggested the possibility that control of the aircraft could have been lost because of a binding or jamming of the elevator. No conclusive evidence of such an event was found in examinations of the wreckage, nor was the source of the reported binding identified.

It is possible that the binding was the result of an unserviceable PTC. The descnption of the binding was similar to that encountered with a previous PTC problem. Under normal circumstances, the PTC is deactivated for take-off, and any irregularities in its operation would not affect take-off. Nevertheless, had it been inadvertently in operation, and malfunctioning, it is remotely possible that abnormal inputs could have occurred as a result of PTC extension. However, had there been such a malfunction, it would be expected that the PTC EXTEND/FAIL light would have been illuminated. Examination of the wreckage determined that the light was not illuminated at impact. Furthermore, extension of the PTC could not explain the significant changes to coefficients of lift and drag.

Testing in the simulator further demonstrated that a runaway PTC during take-off was a situation that was readily overcome by the pilot, resulting in a successful take-off.

Detailed examination of the elevator leading edge revealed the presence of a chordwise scratch on the elevator that corresponded with a mark on the rear spar of the stabilizer. It could not be detemmined if the marks were the result of impact damage or if they existed before the accident. If these marks were not the result of impact, their presence may be indicative of interference between the elevator and stabilizer caused by a foreign object. Such interference could have resulted in the reported binding. Had this been the case, it is also remotely possible that the interference between the elevator and stabilizer progressed to the point that the elevator jammed during the take-off.

Examination of the wreckage detemmined that the elevator was in the full-trailing-edge-up position at impact. Faced with the imminent impact with the terrain, it is likely that the flight crew would have reacted with control inputs that would have resulted in this position. The position of the elevator thus suggests that full-up movement was available to the pilots. Altematively, the impact position of the elevator suggests that, if jamming occurred, it resulted in a full-up deflection, or the jamming was of a transient nature, and the pilots regained authority prior to impact. Had the elevator jammed in the full-up position at rotation, the pitch angle would have exceeded the 8.6 degree geometry limit of the aircraft, and the tail would have struck the runway prior to lift-off. There was no evidence of a tail strike on either the runway or aircraft tail. No scrape marks were observed on the tail skid or on the runway surface. Furthermore, neither of these cases is supported by the analysis of the aircraft's performance during take-off. Neither case would result in the significant changes to coefficients of lift and drag evidenced by the magnitude of the deceleration during the slight climb and the premature stall. Testing in the simulator further demonstrated that jamming of the elevator resulted in pitch angles before lift-off that would result in a tail strike.

Some flight control systems of the aircraft are hydraulically operated. So too are the landing gear and engine thrust reversers. There was evidence to suggest that the hydraulic system of the aircraft was leaking; replenishment of hydraulic fluid was a recurring event. In the two days prior to the accident, a total of 13 quarts of fluid was added to the system. According to Douglas Aircraft Co., leakage is the only explanation that can account for such a fluid replenishment rate. Representatives of the operator suggested that the recorded rate of replenishment had been inflated by vendors and did not reflect the actual replenishment rate. The Board could find no evidence to support this.

Examination of the aircraft wreckage did not reveal any evidence of a hydraulic system failure; however, examination of the hydraulic system was limited to two engine-driven pumps. In view of the significant rate of leakage of hydraulic fluid, it is possible that a hydraulic system failure could have occurred as a result of insufficient fluid. The recovered documentation provided evidence that, on a previous occasion, the pilot had initiated two flights with an inoperative hydraulic system.

Evidence obtained through examination of recovered light bulbs was inconclusive with respect to the status of the main system hydraulic power and services during the take-off. However, the impact status of the hydraulic reservoir low pressure light (not illuminated) would indicate that a rapid depletion of fluid in the main reservoir had not occurred. Furthermore, the impact status of the rudder control manual indicating light (not illuminated) indicates that the rudder was hydraulically powered through either the main system or rudder standby hydraulic power pump.

Testing in the simulator demonstrated that take-off with an inoperative hydraulic system could be accomplished without significant difficulty. Similarly, failure of the hydraulic system during take-off did not result in an unsuccessful take-off. Both the ailerons and rudder automatically revert to manual (aerodynamically boosted) in the event of hydraulic system failure. The horizontal stabilizer is equipped with an alternate electrically powered trim system, and the elevators are operated by a conventional cable system and by an aerodynamic boost tab. The captain's previous take-offs performed with an inoperative hydraulic system further demonstrated that such a take-off could be accomplished without significant difficulty and cannot explain the observed performance degradation during the accident take-off.

EGT indications of the number four engine were approximately 40 degrees hotter than the other three engines. As a result, the Cologne/Cairo sector crew was retarding the throttle slightly on take-off to keep the temperature under limiting values. It is reasonable to assume that the accident crew was doing the same. Information supplied by the engine manufacturer demonstrated that such an action would reduce total engine thrust by about 2.5 per cent. Such an event would have an insignificant effect on take-off performance.

Engines one, two, and three were determined to be operating at high-power settings at ground impact The number four engine was determined to be operating at a lower rpm than the other three engines when it struck the ground. It could not be conclusively determined how much lower the impact rpm was although the position of the bleed valve strongly suggests that, prior to impact with the ground, engine rpm fell below 53 per cent It could not be determined if this lower ground impact rpm was the result of the ingestion of debris as the engine passed through trees immediately prior to ground impact or if the lower rpm was a condition which occurred prior to descent into the trees. With the exception of the possible pre-impact rupture of the pressure regulator diaphragm in the FCU, there was no evidence of any mechanical failure of the engine. Metallization in the transition duct provided positive evidence that the engine was operating at tree impact and had not flamed out.

Independent examination of the number four engine confirmed the assessment of CASB investigators that, with the exception of the possible pre-impact rupture of the pressure regulator diaphragm in the FCU, there was no evidence of any component failure or malfunction involving the number four engine prior to impact with trees and that the engine was operating at the time of tree impact. Similarly, this independent examination could not establish with certainty the engine power output at the time of initial tree impact. However, the independent consultant did conclude that the observed engine damage caused by tree ingestion and resulting deceleration was consistent with a high power output.

It is possible that the rupture of the pressure regulator valve diaphragm of the FCU believed to have been installed on the number four engine occurred prior to impact, although the overall good condition of the diaphragm and previous accident investigation experience suggest that the rupture was impact related and occurred as a result of a pressure spike. Tests with the ruptured diaphragm indicated that, had it occurred prior to impact, no adverse effects would have resulted. However, the FCU bench flow tests were limited to assessing steady state conditions. Thus, the possible effects, such as compressor stalling or surging, a ruptured diaphragm could have had under other conditions, such as a rapid advancement of the throttle lever beyond the take-off thrust position, are not known.

The impact readings of the number one, three, and four engine EPR gauges were consistent with a high power setting. The number two engine EPR gauge reading was consistent with a significantly lower power setting. These gauges are a servo motor type, with no return spring mechanism. Indicators of this type will tend to remain at the position of last reading when electrical power to the system is cut; however, when contacted, the manufacturer of the gauges indicated that, because there is no return spring mechanism, the pointer can move when a gauge is rotated. Thus, it is quite possible that none of the EPR gauges accurately reflected engine power output at impact.

Nevertheless, since three of four EPR gauge impact readings were at or near the take-off thrust setting, their possible significance was examined. In assessing the significance of any individual reading, it is necessary to know when power was removed from the indicator. The impact reading of the number four engine, if reliable, suggests that, when power was removed from the indicator, the engine was operating at high power. Assuming that power was not removed from the indicator until aircraft breakup began to occur, the reading suggests that this engine was operating at high power until at least initial tree impact.

Although the impact reading of the number two engine indicator was well below take-off EPR, it is possible that the reading, if reliable, indicates that power was removed from the indicator later in the impact sequence, after the engine rpm and EPR had decreased as a result of impact and breakup. This assessment is supported by the examination of the engine which indicated that the engine was operating at high rpm at ground impact.

Although there was no definitive evidence to indicate that the number four engine was not operating at a high power setting when the aircraft entered the trees, the possibility that the lower ground impact rpm indicated that an interruption of number four engine power occurred at or after rotation could not be completely ruled out through examination of the engine. Furthermore, witness accounts of the yellow/orange glow could be considered consistent with flames emanating from an engine experiencing compressor stalls and surges. Also considered consistent with an interruption of engine power of the number four engine was the heading change to the right which occurred shortly after lift-off.

Engine performance was not recorded on the FDR. Thus, in the analysis of aircraft performance, it was necessary to assume normal engine operation. Therefore, had there been a power interruption in the number four engine, it could not be distinguished from an increase in drag. However, the thrust penalty associated with the failure of one engine is equivalent to an increase of about 0.05 in the coefficient of drag. The theoretical performance analysis determined that the combined effects of thrust loss or drag increase, necessary to result in the actual performance of the aircraft, were equivalent to a coefficient of drag increase of at least 0.13, well in excess of the value associated with the failure of one engine. Additionally, the failure of one engine cannot explain the significant decrease in coefficient of lift determined in the performance analysis.

Previous accidents involving DC-8 aircraft have demonstrated that, at high angles of attack, it is possible for an engine to experience power fluctuations accompanied by flames emanating from the engine as a result of surging caused by disruptions in the intake airflow. Thus, it is also possible that the lower ground impact rpm of the number four engine and yellow/orange glow observed by witnesses was a consequence of the stall and a subsequent compressor surge that occurred shortly after take-off.

In conclusion, although the possibility of the number four engine operating at less than full power cannot be eliminated, such an event, on its own, should not have caused the accident. Performance simulations conducted on behalf of the Board by UDRI and DND indicated that the performance of the aircraft could be explained by the loss of thrust from one engine, coupled with the performance degradation that results from ice-contaminated wings.

There was evidence to indicate that the potable water system was leaking. Although the system had been subject to maintenance actions in Oakland prior to the initiation of this series of rotation flights, it was again leaking on arrival at McChord, and water leakage was reported by the captain to Arrow dispatch in Miami during a telephone call made from Gander, on the morning of the accident. The Board considered the possible effects that this water leakage could have had on aircraft control either as a result of changes in weight and centre of gravity position or throu~h disruDtion to critical aircraft systems.

Water leaking from the aircraft's potable water system drains by gravity to the space between the cargo compartment liner and the aircraft skin. The lower fuselage is equipped with fuselage drains; however, when the aircraft is pressurized, these drains close and water can accumulate in the belly of the aircraft. During a long duration flight, this water can freeze due to the low ambicnt temperatures at high altitudes. This ice will melt and slowly drain away during ground stops where the ambient temperature is above freezing.

Discussions with other DC-8 operators indicated that, on occasion, water leakage directly into the cargo pits is a problem. The problem is not, however, one of aircraft control, but rather one of wet baggage and water damage to the insulation in the cargo pits. There are no aircraft control systems in the lower portion of the cargo pits which would be affected by water leakage, nor could water accumulate in a quantity sufficient to cause significant changes in the aircraft weight or centre of gravity.

There was no evidence found during the examination of the wreckage to suggest that the aircraft configuration was abnormal at impact.

To assess the position of the flaps at impact, the Board examined evidence gathered through examination of the flap actuators, flap lockout cylinders, flap position indicator, and the flap tracks.

Impact marks inside the flap actuators were consistent with a flap setting of less than 25 degrees. Roller imprints on three of the eight flap tracks recovered were consistent with a flap setting of 18 degrees. Although there were conflicting imprint marks on the other flap tracks recovered, with only two exceptions, these marks were within a corresponding flap setting range of between 12 and 25 degrees. Because of the multiple roller imprints on some flap tracks, the most distinct marks were assumed to be those that occurred at impact. With flaps partially extended, tree contact would tend to pull the flaps and rollers rearward. However, tree contact would not likely produce sufficient shock loading to result in witness marks on the tracks. As a result, witness marks on the tracks could equate to a greater flap angle than the actual position prior to tree impact. Thus, it is possible that secondary impacts occurred during breakup, which may have been of greater magnitude, thus accounting for the range of flap positions determined through interpretation of the most distinct marks. With respect to the remaining two roller imprints, one was clearly unreliable due to the significant difference between imprint positions on the left and right side of the same track (i.e., 50 and 23 degrees). The other imprint which corresponded to a flap position of 32 degrees, was also considered unreliable because of the significant difference in the interpreted flap setting and the flap setting determined for adjacent flap tracks on the same flap.

No useful information was gained through examination of the flap lockout cylinders or the flap position indicator.

Flap asymmetries have been experienced with the DC-8-63. In these cases, the asymmetric condition was caused by failure of a flap-link assembly initiated by fatigue pre-cracking. The flaplink assemblies were recovered from the wreckage and examined. There was no evidence of preimDact failure. No fatigue pre-cracking was detected.

In conclusion, although testing in the simulator demonstrated that severe flap asymmetry could result in a flight profile similar to that of the accident flight, the Board found no evidence to suggest that such an asymmetry had occurred. Based on its examination of the flap system components, the Board concluded that the flaps were extended to the planned 18-degree setting.

The stabilizer angle determined from the wreckage was close to that applicable to the take-off weight and centre of gravity position calculated by the crew and the corresponding V2 speed. It was within the flight-deck indicator's 1 ANU margin of error. Because of indications that the flight crew had underestimated the take-off weight and may have inadvertently used a V2 speed applicable to 310,000 pounds, the corresponding take-off stabilizer angle was calculated. This value (5.8 ANU) was also close to the value determined from the wreckage. It too was within the flight-deck indicator's 1 ANU margin of error. Thus, the Board concludes that an inappropriate stabilizer setting did not contribute to this accident.

Examination of the recovered wing slot hydraulic actuators suggested that the wing slot doors were in the appropriate (open) position at impact This conclusion was supported by the determination that the wing slot door light was not illuminated at impact. This light will illuminate when the wing flaps are not in the UP position and any one or none of the slot doors is not fully open.

The results of the performance analysis and simulator testing further indicated that closed slots could not explain the accident. The lift penalty which results from closed slots is a 0.2 reduction in maximum coefficient of lift. The performance analyses calculated that a minimum 0.38 decrease in maximum coefficient of lift is necessary to result in an increase in stall speed of the magnitude indicated through analysis of the FDR recording. Testing in the simulator demonstrated that take-off with wing slots closed could be completed without sigrnificant difficulty.

There was no evidence to suggest that an inadvertent extension of the ground spoilers had occurred. Examination of the ground spoiler system hydraulic actuator determined that it was in the extended position at impact, consistent with spoilers retracted. The lift and drag penalties associated with their deployment exceed the values determined in the performance analysis. Although the Board was unable to successfully simulate the in-flight deployment of the ground spoilers, it has no doubt that such an event, if it were to occur immediately after take-off, would result in catastrophic consequences not dissimilar to those which occurred on the morning of 12 December 1985. Nevertheless, there was no physical evidence to suggest that such an event had occurred. Furthermore, the operation of the spoiler system through a ground shift mechanism and nose gear oleo extension prevents the spoiler lever from being inadvertently moved to the EXTEND position when the aircraft is in the air.

The landing gear was extended at impact. Normally, retraction of the landing gear is initiated within three seconds of lift-off, once a positive climb rate has been established. In view of the severely degraded climb performance after lift-off and the abnormal flight characteristics associated with the stall onset, night management problems likely precluded an up selection of the landing gear. Tests in the simulator confirmed that, when faced with a situation involving degraded climb performance, a gear-up selection was rarely completed.

Initial examination of the number four thrust reverser at the accident site raised the possibility that the reverser had deployed in night. When found, the translating ring of the reverser system had been turned inside out, giving the appearance that the reverser had been open at ground impact. This possibility was further supported by the aircraft's slight turn to the right shortly after lift-off. As a result, all four engine thrust reversers were subjected to close scrutiny by investigators. In the case of engines one, three and four, the translating rings were determined to be in the forward position and the deflector doors faired. In the case of the number two engine, the translating ring may have been aft of the forward stop but was at least some 16 inches forward of the rear stop and the deflector doors were faired. The Board considers this to be clear physical evidence that all four reverser assemblies were in the forward thrust position at impact.

No pre-impact faults with the reversers were identified.

Consideration was given to the possibility that a reverser had deployed in flight and, as a result of crew actions, had been stowed prior to impact. The performance penalties associated with deployment in flight are considerable. Simulator testing showed that application of full reverse thrust on the number four engine at or near lift-off could result in a flight profile similar to that of the accident flight.

The aircraft is equipped with an emergency "dump" capacity which, when selected, instantly returns the reverser doors to the faired position, thus eliminating reverse thrust. In the accident aircraft, the emergency dump switch was located on the overhead console above the captain's (left-hand) seat. The dump switch can not, however, move the translating ring forward to the stowed position. Thus, if a reverser had deployed in flight and the dump switch activated, only the doors would fair and the translating ring would have remained in the aft position.

Therefore, when the position of all four reverser assembly translating rings is considered, uncommanded deployment of a thrust reverser could not have occurred.

There was considerable speculation that the accident occurred as a result of the detonation, either accidental or through sabotage, of some explosive device. This speculation was fuelled by the fact that military personnel and equipment were aboard the night and by the increasing worldwide incidence sof terrorist activity. Also contributing to this speculation were a reported claim of responsibility by a terrorist group, the point of origin of the flight, and the reports by three witnesses of a yellow/orange glow emanating from the lower surface of the aircraft. The observations of the yellow/orange glow also raised the possibility of a pre-impact fire.

Detailed examination of the wreckage with the assistance of forensic experts of the RCMP, including examinations at the RCMP Central Forensic Laboratory, revealed no evidence of an explosion or pre-impact fire. All damage to the aircraft and its components was considered to be the result of impact with terrain and the post-crash fire.

The Board believes there is sufficient evidence to conclude that two side panels were missing in the number three cargo pit. The absence of these panels would compromise the integrity of the Class D classification of this compartment. A Class D cargo or baggage compartment is one in which: a fire occurring in it will be completely confined without endangering the safety of the airplane or occupants; there are means to exclude hazardous quantities of smoke, flames, or other noxious gases from any compartment occupied by the crew or passengers; and ventilation and drafts are controlled within each compartment so that any fire likely to occur in the compartment will not progress beyond safe limits. Thus, although the Board found no evidence to suggest that a fire had occurred in the number three cargo pit, the missing side panels would permit ventilation of the compartment and. in turn, possible propagation of a fire, if one had originated in this compartment.

Examination of the engine fire extinguishing agent containers indicated that it was possible that agent had been released into the number three engine as a result of crew actions; the explosive charge had fired while agent was still in the container. This and witness observations of the yellow-orange glow raised the possibility of a pre-impact fire in the number three engine. However, other evidence indicates that this did not occur. Intentional discharge of the fire extinguishing agent into an engine through operation of the fire extinguishing agent discharge switches first requires movement of the appropriate engine fire shut-off lever. One of the functions of this lever is to close the fuel shut-off valve, thereby shutting down the engine. The number three engine was determined to be operating and at high rpm at ground impact. This indicates that the engine had not been shut down prior to ground impact The evidence also indicates that the Master Fire Warning light was not illuminated at impact. Activation of the fire extinguisher would also be contrary to Arrow Air published emergency procedures and training which specify that, in the event of an emergency during take-off, flight crews are to wait until a safe altitude (1,000 feet AFE) is attained before dealing with specific problems.

Discussions with the manufacturer of the fire extinguishing agent containers indicated that it is possible for the explosive cartridge in the container to activate as a result of exposure to the high temperatures associated with a post-crash fire or through energizing of the actuating circuit during aircraft breakup. In consideration of all of the available evidence, the Board concludes that the discharge of the fire extinguishing agent was the result of either impact or the post-crash fire and not the result of an intentional action on the part of the flight crew.

Despite an extensive search of the area between the departure end of the runway and the initial impact point, no components or debris was found that originated from the aircraft evidence that the aircraft was intact until initial impact with the terrain.

There was no evidence found of any ammunition or military ordnance in the wreckage. A thorough inspection of personal baggage loaded on board the aircraft had been carried out prior to departure from Cairo. No explosive materials or otherwise hazardous items were discovered. The Board noted no significant difference between the weapons recovered and those reported to have been on board.

Several small post-impact explosions occurred in the burning wreckage. Although some of these explosions were reportedly large enough to cause mounds of rubble to lift several feet into the air, none were considered of sufficient magnitude to be the result of detonation of explosive devices. The Board attributes these explosions to the normal bursting of pressure vessels (accumulators, fire extinguishers, aerosol cans, etc.) due to the heat of the fire. It is also likely that some of the reported explosions may have been firing of up to ten .45 calibre small arms rounds reported to have been carried on the aircraft by the Batallion Commander and the CID inspector.

The occurrence of a pre-impact fire or explosion was also not supported by the autopsy evidence and the blood carboxyhemoglobin levels of the aircraft occupants.

No evidence was found of shrapnel wounds and/or the identifiable portions of an explosive device, nor were injury patterns deemed to be characteristic of a pre-impact explosion.

All of the pathologists involved in the assessment of the pathological/toxicological evidence agreed that pathological examinations and toxicological analyses yielded no evidence of pre-impact inhalation of the products of combustion and that, when these findings were combined with evidence from the accident site, injury patterns and mechanisms and timings of death, pre-impact inhalation of products of combustion could be excluded beyond any reasonable doubt.

Although there was some level of HCN detected in the remains of the majority of aircraft occupants, it was the conclusion of all pathologists involved in the assessment of the pathological and toxicological findings that the HCN values were unreliable as an indicator of pre-impact fire and, at best, only indicative of exposure to fire. A high correlation with exposed chest cavities and hemothorax was noted in the cases with very high HCN concentration. In the 20 cases with the highest HCN concentration, 17 cases had exposed chest cavities and 16 had either documented hemothorax or multiple rib fractures which was accepted as evidence of hemothorax. This represented a highly significant correlation between high HCN levels and hemothorax. Almost all the blood samples were retrieved from the body cavities, and, thus, it was the agreement of all pathologists involved that much of the HCN in the blood was the result of post-mortem exposure to fire. The effects of neo-formation on the HCN levels, if any, could not be identified.

CO values were considered to be a reliable indicator of the inhalation of the products of combustion. In this regard, all cases of elevated CO levels were considered to be the result of postimpact inhalation of the products of combustion.

In summary, it was concluded that all aircraft occupants died as a direct result of impact and/or the post-crash fire. Some of the victims sustained injuries compatible with short-term survival and died as a result of inhalation of the products of combustion, either primarily or in combination with severe injuries sustained during impact. No evidence of any pre-impact fire or explosion was found as a result of the pathological examinations and toxicological testing.

Finally, the performance of the aircraft was not consistent with a sudden and catastrophic event such as an explosion.

Considerable interest was generated by the yellow/orange glow reported by some witnesses. However, in the absence of corroborating physical evidence, the Board was unable to determine the source of the illumination described by these witnesses. In assessing the significance of this evidence, the Board took into account that each saw the aircraft for only a brief period of time, and, since all were driving vehicles when they made their observations, they could not fully direct their attention to the aircraft. As a result, none was able to precisely describe the phenomenon, nor fix its position on the aircraft. Although at least one of these witnesses thought that the glow might have been a fire, he was not certain. Experience has shown that, when an accident is followed by a post-impact fire. witnesses often tend to associate fire with pre-impact observations.

The Board also noted that other witnesses who observed the aircraft during its brief flight did not report observing this glow or any other observation consistent with a fire. Two of these witnesses observed the take-off of the aircraft until after it began to descend below trees beyond the departure end of the runway.

It is possible that the glow observed by some witnesses was the illumination from normal light sources on the aircraft such as landing lights. One of these witnesses attributed the phenomenon to the reflection, on the bottom of the aircraft, of approach lights for runway 04 located on the extended centre line of runway 22. It could not be determined if the approach lights to runway 04 were illuminated at the time of the accident. It is also possible that the phenomenon observed by these witnesses was caused by compressor surging of one or more engines, resulting from disruptions in intake airflow. Compressor surges accompanied by flame emanating from the engine have been observed in other DC-8 accidents where angles of attack at or beyond the stall were achieved.