Normal take-off procedures are described in the Arrow Air Inc. DC-8 Airplane Operating Manual. Pertinent extracts from the manual follow:

1. "With a smooth positive back pressure, initiate rotation of the airplane at the scheduled VR speed. Adjust the rate of rotation [of 2 degrees per second. Do not allow the pitch attitude on the runway to exceed] maxlmum 8 degrees, so as to attain the V2 speed at a height of 35 feet above the runway surface." *

2. "Retract gear as soon as a definite climb is established...."

3. "After gear is up accelerate to V₂ - 10."

4. "WARNING: Failure to remove snow and ice accumulated on aircraft while on the ground can result in serious aerodynamic disturbances and structural damage when flight is attempted. Take-off distance and climb-out performance can be adversely affected to a dangerous degree, depending on weight and distribution of the snow and ice. Structural damage has also resulted from vibrations induced in flight by unbalanced loads of unremoved accumulations. These hazards must be avoided by removing the snow and ice from the wings, fuselage and tail before flight is attempted."

Take-off With Engine Failure

The Arrow Air Inc. DC-8 GroundFlight Training Manual, under the general heading "Take-off with Engine Failure" states the following:

The Arrow Air Inc. DC-8 Airplane Operating Manual, under the general heading "DC-8-63 Cold Weather Procedures" states the following:

The Weight and Balance section of the Arrow Air Inc. DC-8 Operating Manual identifies the standard average adult passenger weight, including five pounds of carry-on baggage, for use between 01 November and 30 April as 170 pounds. This section also states that actual passenger weights should be used when large groups of passengers are carried whose average weight does not conform to the normal standard weight. Examples are given as a group of large athletes or a planeload of men.

On 31 October 1985, Arrow Air published Bulletin 85-22 which introduced a new system for calculating the weight and centre of gravity position of its DC-8 aircraft. This new system, entitled Arrow Air's Adjusted Weight Unit Loading System, was designed to simplify and give greater accuracy to the development of the loading system analysis prior to the dispatch of each flight.

Instructions for the operational use of the loading system were contained in the bulletin. To determine passenger weight, flight crews were instructed to enter on the load sheet the number of passengers to be boarded and then enter the adjusted weight units from the loading table which corresponded to the number of passengers. The weight units found in the loading tables were based on an average passenger weight of 165 pounds in summer and 170 pounds in winter. There were no instructions or guidance concerning the requirement or method to determine total passenger weight using actual passenger weights when the average value was not considered representative of actual passenger weights.

After Bulletin 85-22 had been developed, the FAA principal operations inspector (POI) assigned to Arrow Air was consulted. and he concurred with its contents.

The Douglas Aircraft DC-8-63 Weight and Balance Manual defines operational empty weight as the basic empty weight plus operational items. Operational items are identified as those items of personnel equipment and supplies that are necessary on a particular operation unless already included in the basic empty weight. Examples of items normally included in the operational empty weight are flight crew, removable cabin and galley equipment, and usable drinking and washing water.

The Arrow Air weight derivation report for N950JW indicated that the aircraft was last weighed on 04 August 1985. The basic empty weight was determined to be 159,399 pounds. Examina- tion of the pre-weighing check-list determined that this weight did not include removable galley equipment, cabin items such as pillows and blankets, or disposable water. The derivation of operational empty weight (basic operating weight) included only the weight of the flight crew and their personal baggage.

Although the weight of the flight attendants and meals was included in the determination of the aircraft weight and centre of gravity, no consideration was given to removable galley equipment, removable cabin items, and potable water.

By contrast, the previous weight derivation performed by Union des Transports Aériens in 1981 included 1,250 pounds in the basic operating weight to account for these items.

In consideration of the above, the Board estimates that the basic operating weight of the aircraft was underestimated by at least 1,000 pounds.

The Douglas Aircraft DC-8-63 Weight and Balance Manual defines maximum design ZFW as "the maximum weight of an aircraft less the weight of all usable fuel and other consumable propulsive agents in particular sections of the aircraft that are limited structurally to this condition. This is a weight at which the subsequent addition of fuel and other consumable propulsive agents (as limited by other design gross weights) will not exceed the aircraft design strength." The Douglas Aircraft DC-8-63 Weight and Balance Manual states that the actual ZFW must never exceed the maximum design ZFW.

The maximum design ZFW of N9SOJW was 230,000 pounds. In 1985, Arrow Air explored the possibility of increasing the maximum design ZFW of the aircraft. A Supplementary Type Certificate (STC) was available that would have raised the maximum design ZFW of the aircraft by 14,000 pounds. No structural modifications to the aircraft were required; however, certain modifications to the airspeed indicating system were necessary to provide maximum airspeed warnings for a reduced flight envelope. The maximum allowable airspeed would have been reduced, but this speed, 352 KIAS versus 373 KIAS, would not normally be exceeded during normal operation. Although action to obtain the STC had been contemplated, it was not being actively pursued.

Weight and balance calculations were performed by the crew using the Arrow Air Adjusted Weight Units Loading System for DC-8-62 and DC-8-63 passenger aircraft.

The load sheet completed at Gander showed a 330,625 pound gross take-off weight which was comprised of the following:

The calculated centre of gravity position was 25.4 per cent Mean Aerodynamic Chord (MAC), well within the allowable range of 14.0 to 32.3 per cent MAC.

With the exception of the take-off fuel weight, all weight values had been determined at Cairo by the Cairo/Cologne flight sector crew. Because the passenger and cargo loads did not change at Cologne, these same values were used for the subsequent take-offs from Cologne and Gander. Weight and balance calculations in Cairo were performed by the first officer.

The passenger weight used was the winter-adjusted weight unit which corresponded with 250 passengers. This represented an average passenger weight of 170 pounds. Although the load had changed in Cairo, the cargo pit weights used were the same weights used to perform weight and balance calculations for the flights from McChord AFB to Cairo. The first officer and flight engineer who operated the flight testified at the Board's public inquiry that they believed the actual take-off weight at Cairo to be about 10,000 pounds greater than that calculated on the load sheet. Although they increased take-off reference speeds accordingly, the calculations on the load sheet were not amended, nor was this information passed to the crew who assumed responsibility for the aircraft at Cologne. The captain of the aircraft on the Cologne/Cairo flight sector testified that he did not recall either the first officer or flight engineer informing him that take-off reference speeds had been increased.

Weight and balance calculations for the first series of rotation flights on 03 to 05 December were reviewed by investigators. These calculations were performed using the same adjusted weight units loading system by the same flight crew who performed the calculations for the 10 to 12 December series of flights.

The passenger weight indicated on the load sheets for this first series of flights was 8,000 pounds greater than that used on the 10 to 12 December flights. The total weight of cargo indicated was 8,000 pounds less than that of the 10 to 12 December flights. The weight used on the flight to Cairo was identical to those used on the flights to Fort Campbell.

In preparation for the 12 December flight from Cairo to Fort Campbell, all cargo which was to be placed in the aircraft cargo pits was weighed. The weighing was performed by MFO personnel, prior to departure from their base of operations in the Sinai. The weight of the cargo was determined to be 27,950 pounds and consisted of 481 duffle bags and 48 foot lockers of miscellaneous military goods.

At Cairo, it proved impossible to fit all the cargo into the aircraft. Despite extraordinary efforts and to the expressed consternation of MFO personnel, 41 duffle bags were left behind in Cairo. After the accident, MFO personnel estimated the weight of the items left behind to be 2,870 pounds. This figure was determined by MFO personnel who estimated the average weight of each duffle bag to be 70 pounds.

It could not be determined if the scaled weight determined by the MFO was passed to Arrow Air personnel at Cairo. The total cargo weight indicated on the load sheet was 2,400 pounds less than the scaled weight determined by MFO personnel. The cargo weight indicated on the load sheet also included 1,300 pounds of catering equipment and aircraft spares not included in the MFO scaled weight. Thus, even if the estimated 2,870 pounds of duffle bags which were not loaded on the aircraft were considered by the crew, the cargo weight was about 1,000 pounds heavier than that indicated on the load sheet.

The weight of the passengers was not determined on departure from Cairo. For flight planning purposes, an average weight of 170 pounds was used to determine the total passenger weight. This average weight includes an allowance of five pounds for carry-on baggage. Information from several sources indicates that the total weight of passengers and their carry-on baggage was considerably in excess of the weight calculated by the crew using the 170-pound average value. Ante-mortem weights of the 248 passengers were determined through the examination of U.S. army personnel records. The average weight of each passenger without uniform was 164 pounds. Each passenger carried with him personal gear which included a weapon, miscellaneous military equipment, web belt, clothing, and souvenirs. The carry-on baggage boarded on the aircraft in Cairo nearly filled the baggage holds of two B-737 aircraft which ferried the passengers from their base at Ras Nasrani to Cairo. Numerous witnesses indicated that a large quantity of carry-on baggage was stowed in the cabin on departure from Cairo. The quantity of cabin baggage was the subject of concern to MFO personnel and the cabin crew.

In an effort to determine the total weight of passengers and cabin baggage, the passenger weights of MFO flights inbound to Cairo were examined. In accordance with procedures established by the U.S. Army, each passenger travelling from McChord AFB to Cairo for duty with the MFO was weighed with his or her carry-on baggage.

On 03 December 1985, the total scaled weight including carry-on baggage of the passengers who flew from McChord AFB to Cairo was 54,726 pounds or about 219 pounds per passenger. This value was passed to Arrow Air personnel at McChord. Personnel who dealt with the flight in bound to Cairo on 10 December and the flight outbound to Fort Campbell on 12 December reported that the cabin baggage on the outbound flight exceeded that on the inbound flight.

In consideration of the points enumerated above, the Board estimates that the average weight of each passenger on the accident flight was about 220 pounds. The total weight of the 248 passengers was thus about 54,560 pounds, that is, about 12,000 pounds higher than the weight indicated on the load sheet.

Based on the findings of its investigation, the CASB estimated the take-off weight of the aircraft to be about 344,500 pounds comprised of the following:

The movement of MFO troops to and from Egypt was governed by a contract between Arrow Air and the MFO. entered into in 1984 and renewed in 1985.

The contract specified that, for each flight, 250 passengers were to be carried and each passenger was entitled to a baggage allowance of 154 pounds. No passenger weight was specified in the contract, nor was there any requirement for the MFO to provide weight information to Arrow Air for individual flights.

In preparation for this series of flights, planning meetings were held involving U.S. Army personnel and representatives of Arrow Air. At those meetings, Arrow Air representatives informed U.S. Army personnel that the payload capacity of the DC-8 was 72,000 pounds. It was the consensus among all concerned that, on these flights, the aircraft would "bulk-out" before the maximum weight capacity was reached.

Various directions and instructions were proposed by the U.S. Army pertaining to the movement of their personnel to MFO duties in the Sinai. A standard operating procedure, promulgated by the U.S. Army for the use of units deploying to duty with the MFO, identified the payload capacity of the deployment aircraft as 75,000 pounds. The individual baggage allowance established by the U.S. Army for MFO members was 150 pounds. For certain personnel, the baggage allowance was 175 pounds. For planning purposes, the U.S. Army considered 170 pounds to be the average weight of each soldier. When planning for a tactical deployment, 220 pounds was used as an average weight to account for web gear and weapons.

A U.S. Military Command pamphlet designed for use by U.S. military organizations when planning airlift requirements identified the allowable payload of a DC-8-63 as 90,000 pounds. No manifests which contained weight information were prepared by MFO or U.S. Army personnel either at Cairo or at McChord AFB, nor were they requested to do so by Arrow Air. In all cases, where weight information was passed on, it was done via miscellaneous slips of paper.

According to an operation order prepared by U.S. Army personnel which described the procedures governing this deployment of troops to the Sinai, the officer-in-charge was to have on paper the total weight of all personnel and baggage to be loaded on the aircraft. Three copies were to be prepared. One was to go to the aircraft captain and one to the MFO representative. The only documents recovered during the investigation pertained to the weight of the baggage boarded at Cairo.

The only U.S. military records recovered which pertained to the payload carried on this series of MFO rotation flights were records and audit documents prepared by U.S. Air Force personnel at McChord AFB to account for the use of ramp space and ground equipment at McChord.

The recorded payload on 03 December was 40,000 pounds passengers and 14,760 pounds cargo. On 10 December, the recorded payload was 50,000 pounds passengers and 37,500 pounds cargo.

The only reference to the aircraft load found in Arrow Air documents was in the flight message addressed to the flight crew in Cologne from Arrow Air dispatch personnel in Miami. The message contained a note to plan for 250 passengers with 100 pounds of baggage each.

The original passenger load planned for this rotation flight was 250. However, in the days immediately preceding the flight, several adjustments to the passenger manifest were made, and the planned load was reduced to 249 passengers. The actual passenger load was reduced to 248 because one passenger, who was to have been on board the aircraft, misplaced his passport and was not permitted to board the aircraft at Cairo. However, his personal baggage remained on the aircraft. The load sheet prepared by the flight crew in Cairo and carried over to the departures from Cologne and Gander showed a load of 250 passengers.

In addition to their own personal effects, the military personnel on board the aircraft carried personal issue military equipment which included a variety of weapon types. The United States Army provided the following list of weapons believed to be on board the aircraft at the time of the accident:

Other miscellaneous military equipment belonging to the unit was also on board the aircraft. However, with the exception of one clip each of .45 calibre ammunition reported to have been carried by a Criminal Investigation Division (CID) inspector and the Batallion Commander, this equipment did not include military ordnance, ammunition, or other explosive material.

All personal effects carried on board the aircraft were subject to a rigorous pre-flight inspection by United States Military Customs Inspectors and Egyptian Customs officials. Approximately 60 per cent of the baggage placed in the cargo pits of the aircraft was inspected. Bags were selected at random, emptied and the contents examined. One hundred per cent of the carry-on baggage was inspected. No unauthorized military equipment, ordnance, explosives, or military devices of any kind were found during this inspection procedure, nor had any such items been found in similar inspections conducted prior to the first flight of this rotation and prior to the three flights in the preceding rotation.

The baggage that belonged to the passenger who was not boarded at Cairo had been subjected to this inspection procedure.

The FAA approved DC-8-63 Airplane Flight Manual defines the minimum take-off field length as the greatest of:

1. The distance from start of takeoff to a point 35 feet above the runway at the V₂ speed, assuming an engine to fail at a speed corresponding to the decision speed, V₁.

2. The distance to accelerate to the decision speed, V₁, and to bring the airplane to a stop. The stopping performance is based on maximum braking on a dry, hard surface runway, anti-skid operative, with spoiler extension initiated after the throttles are moved to the idle position.

3. The all-engines-operating takeoff field length which is 115% of the four-engine distance from start of takeoff to the 35 foot height.

The take-off run available was 9,900 feet because runway 22 was entered from a right turn from runway 13. The field-length limited take-off weight for the accident night, as determined from the Flight Manual, was 352,000 pounds.

The following take-off reference speeds are utilized in DC-8 flight operations:

1. V₁ - Critical Engine Failure Speed.

2. V_R - Take-off Rotation Speed. The speed at which rotation is initiated during the take-off to achieve the V₂ climb speed at 35 feet.

3. V₂ - Take-off Climb Speed. The V₂ value is equal to the actual speed at the 35-foot height, as demonstrated in flight tests and must be equal or greater than 120 per cent of the stall speed.

4. V_F - Flap Retraction Speed. The minimum flap retraction speed. It is equal to V₂ + 25 knots.

The take-off reference speeds are normally determined by the flight engineer and by reference to tables found in the DC-8-63 Airplane Flight Manual. They are then reviewed by the captain and first officer and are set using movable "bugs" located in the circumferential ring of each pilot's airspeed indicator, with the exception of the V₂ value which is set using a rotary knob which moves a cursor behind the glass face of each indicator.

Take-off reference speeds and corresponding stabilizer angles vary depending on the take-off weight and centre of gravity position of the aircraft and the flap setting used for take-off. The take-off reference speed data card calculated by the crew for the take-off at Gander was not found. Applicable take-off reference speeds and corresponding stabilizer angles for take-off weights of 310,000, 330,600, 344,500 and 355,000 pounds and the applicable centre of gravity position, as published in the DC-8-63 Airplane Flight Manual, are as follows:

Corresponding stabilizer angle for take-off:

The 172 knots indicated on the captain's airspeed indicator internal bug did not correspond with any V₂ value published in the DC-8-63 Airplane Flight Manual.

Based on a take-off weight of 344,500 pounds (estimated by the Board) and a V₂ speed of 158 KIAS (V₂ that corresponded with the internal bug setting on the co-pilot's airspeed indicator) the corresponding stabilizer angle for take-off would be 5.8 ANU.

Douglas Aircraft Co. supplied a considerable amount of information pertaining to the aerodynamic performance of a DC-8-63. This performance information took into account the aircraft configuration and ambient conditions at Gander and included data for a normal take-off and data applicable to certain abnormal conditions.

According to information supplied by Douglas Aircraft Co., under the ambient conditions at Gander, a DC-8-63 with a take-off weight representative of that estimated by the Board should have lifted off 47 seconds after brake release at 165 KIAS, following a ground roll of 6,700 feet, assuming that rotation was initiated at 153 KIAS and the time from rotation to lift-off was 3.5 seconds. The take-off distance (to 35 feet agl) would have been 7,800 feet. An engine failure at V_R would have increased the take-off distance by about 200 feet.

Further information was supplied which pertained to changes in the coefficient of lift generated by the lift-producing surfaces of the aircraft under the following conditions: leading edge slots closed; ground spoilers deployed; and ice-contaminated wings.

The lift penalty which results from closed wing slot doors is a 0.2 reduction in maximum coefficient of lift. Ground spoiler deployment results in a 0.4 reduction in coefficient of lift at zero degrees angle of attack. This reduction increases as angle of attack increases. Douglas Aircraft Co. was unable to provide exact data for higher angles of attack. With respect to ice contamination, Douglas Aircraft Co. supplied information which indicated that, with wings contaminated by surface roughness elements of 0.04 inches, maximum coefficient of lift would be reduced by about 25 per cent. The coefficient of drag at or beyond the stall angle of attack (which would be a lower than normal angle) would increase by greater than 100 per cent relative to an uncontaminated wing operating at the same angle of attack, below the stall. (See Figure 1.17.)

An aircraft will not lift-off until the lift produced exceeds the aircraft weight. Because the DC-8-63 is geometry-limited to a pitch angle of approximately 8.6 degrees on the ground, crews have been trained not to allow the pitch angle to exceed eight degrees while the aircraft is still on the ground. Thus, for the DC-8-63 to become airborne, the aircraft must accelerate to a speed where sufficient lift will be generated at the limiting angle of attack.

Experience has shown that the DC-8-63 begins to rotate approximately one to two seconds after the "rotate" call is made, assuming normal crew and aircraft response times. If the aircraft is rotated early but at a normal rate of two degrees per second, the DC-8-63 will reach a pitch angle of eight degrees before reaching a speed that will produce sufficient speed for lift-off. If the eight-degree pitch angle is held, lift-off would occur at about 161 KIAS at an aircraft weight representative of that estimated by the Board.

For small angles of climb at a given aircraft weight, the rate of climb depends on the difference between thrust and drag. When the total thrust is greater than the total drag, the aircraft is able to climb at a steady or increasing speed. When the aircraft climbs at an angle greater than al- lowed by the available excess thrust, the airspeed will decrease.

If airspeed, climb gradient, and thrust are known, the total aircraft drag can be calculated for the climb after lift-off. The total drag can be used to calculate a coefficient of drag required to produce the climb profile. Using FDR data, the coefficient of drag required to produce the climb profile of the accident flight was calculated. In the absence of reliable FDR altitude data, maximum and minimum climb profiles for the accident flight were determined using witness observations and the JETS Mode C altitude readout. In this manner, the peak altitude achieved during the brief climb after take-off was determined to be no more than 125 feet above the runway. Similarly, thrust was assumed to be normal four-engine take-off thrust for the ambient conditions at Gander.

From 54 seconds to 61 seconds after brake release, the airspeed decreased at a rate of 1.3 knots per second. Three calculations were performed to assess the coefficient of drag necessary to ac- count for the observed deceleration. These calculations assumed altitude gains after take-off of 70 feet, 100 feet and 125 feet respectively. The results of these "snapshot" calculations suggested that coefficients of drag of 0.29, 0.281, and 0.267 would be required to explain the performance, assuming that all four engines were developing take-off thrust and respective altitude gains of 70 feet, 100 feet, and 125 feet occurred.

The loss of thrust from one engine is equivalent to a change of 0.05 in coefficient of drag. Thus, if an engine failure is considered in these calculations, the coefficients of drag that would be required to explain the performance of the aircraft are 0.24, 0.231, and 0.217 respectively.

The manufacturer's data indicated that the expected coefficient of drag would be approximately 0.13 for a normal climb following lift-off.

The lift produced by an airfoil (wing) is primarily dependent on three variables: airfoil geometry, angle of attack, and airspeed. Airfoil geometry on any given aircraft is altered through the use of trailing edge and/or leading edge flaps. Typically, extension of flaps increases the lift-producing capability (coefficient of lift) of a wing. For a specific flap setting, the only other way of changing the coefficient of lift is to change the angle of attack.

Angle of attack is the relative angle between the air impinging on the wing and the wing chord. As the angle of attack increases so does the coefficient of lift. The coefficient of lift continues to increase with increases in angle of attack as long as the airflow over the wing remains smooth and adheres to the contour surface of the wing. However, at a certain angle of attack, the airflow begins to separate from the upper surface of the wing. Initial separation usually occurs near the trailing edge of the wing. As the angle of attack increases further, the separation points move forward until the critical angle of attack is reached. Beyond this critical angle, any further increase in angle of attack results in a decrease in coefficient of lift, and a stall is said to have occurred. Near the stall, drag increases significantly.

The point at which an aircraft will stall is dependent upon angle of attack. However, due to the interrelation of angle of attack and airspeed in the production of lift, stalling and the point at which an aircraft will stall are usually expressed in terms of airspeed. For a given flap angle, factors which affect stall speed are thrust, angle of bank, load factor (vertical acceleration, 'G'), weight, and centre of gravity position.

Stall onset is the flight regime that precedes a full stall. In this regime, the aircraft is subjected to ever increasing buffet, pitch, and roll activity. Typical stages of stall onset, in order of occurrence, include activation of the artificial stall warning; momentary separation of the airflow on the wing as airspeed is reduced toward stall; buffeting which increases in intensity as speed decreases further (and angle of attack increases); the movement forward of the centre of lift as the separated flow region expands, resulting in less pilot control force necessary to cause the nose to raise; an increase in roll activity and lateral control difficulties caused by asymmetries in the fluctuating separation regions of each wing and which typically result in aircraft heading change during stall.

The stall characteristics of the DC-8 series of aircraft are described in the DCA Flight Study Guide in the following manner. "The stall characteristics of all DC-8 series aircraft are excellent and straightforward in every respect throughout the entire operating weight and C.G. range. All aircraft possess a crisp, clean break with no pitch-up tendencies or adverse roll characteristics. This is basically achieved through care in wing design. On the DC-8, as angle of attack is in- creased in the approach to stall, the inboard section of the wing, which already has been flying at a greater angle of attack than the outboard because of airfoil design, will stall first. The center of lift of the wing will then move aft with respect to the aircraft's center of gravity, thus causing the nose to pitch down in a positive manner while good lateral control is retained."

According to aerodynamicists from Douglas Aircraft and published material from the Boeing Commercial Aircraft Company (Jet Transport Performance Methods), swept wing jet transport aircraft like the DC-8-63 will yaw at the stall, particularly if pilots are trying to control wing drop with aileron inputs. An examination by the Board of several other accidents in which a DC-8 aircraft stalled shortly after lift-off determined that it was common for the aircraft heading to deviate significantly when the aircraft stalled.

Stall speeds for the DC-8-63 are published in the FAA approved Airplane Flight Manual. The speeds are predicated on idle thrust and a forward centre of gravity. They represent the minimum speed reached during aircraft certification stall recovery tests. With 18 degrees of flap and an aircraft gross weight of 344,500 pounds, the FAA certified stall speed is 144 KIAS.

Actual stall speed, that is, the speed at which the stall occurs, is higher than the FAA certified stall speed. Calculations using aerodynamic data provided by Douglas Aircraft Co. produced a 1G stall speed for a gross weight of 344,500 pounds that was about 11 knots higher than the FAA stall speed. A centre of gravity position which corresponded with the take-off from Gander would result in a stall speed decrease of three knots. Take-off thrust would reduce the stall speed by a further four knots. Because of the limitations of the FDR, the effects of load factor and angle of bank could not be estimated. Accordingly, the 1G clean wing stall speed on take-off from Gander, determined from data provided by Douglas Aircraft Co., was about 148 KIAS.

Most jet transport aircraft are equipped with a stick shaker or some other type of warning device to alert the pilot that the aircraft is approaching a stall. In the DC-8, the stick shaker is activated by a sensing mechanism (a lift transducer) in the wing leading edge.

The vane of the lift transducer protrudes through the lower surface of the wing leading edge so that, during flight, aerodynamic forces on the vane activate the stick shaker when a preset angle of attack is reached. According to Douglas Aircraft Co., the DC-8-63 stick shaker will activate approximately 13 knots above the FAA stall speed and six knots above the 1G stall speed.

Any airplane operating near the ground will experience changes in the aerodynamic characteristics of its wing. The ground will cause a restriction to the local airflow and alter the wing upwash, downwash, and tip vortices. These effects are referred to as ground effect.

In ground effect, the induced flow velocities will be reduced, and the wing will experience a lower induced drag coefficient and a higher coefficient of lift for any specific angle of attack. In other words, the wing will require a lower angle of attack to produce the same lift coefficient and the corresponding drag coefficient will also be lower.

Ground effect is most pronounced when the aircraft is within one quarter of a wing span (37 feet for the DC-8-63) of the ground. All ground effect benefits are lost when the aircraft is more than one wing span above the ground. Typically, induced drag is reduced by about 20 per cent at one-quarter wing span from the ground and by about 45 per cent at one-tenth wing span from the ground.

Considerable research has been conducted into the effects of ice contamination on airfoil performance. As a result of this research, it is accepted that, in general, ice accretion on the leading edge or upper surface of an aircraft wing results in an increase in stalling speed, decrease in the stall angle of attack. and rapid drag increase near the stall. (See Figure 1.15.)

These effects and the inherent hazards have been documented and described in numerous aerodynamic tests and papers, aviation periodicals, and cold weather operations manuals. (See Appendix C for general information on the aerodynamic effects of ice contamination).

Recent research has shown that seemingly insignificant amounts of wing ice can be sufficient to significantly degrade an aircraft's performance and flight characteristics. Surface roughness caused by ice, frost, snow, or even large accumulations of insect debris or badly chipped paint can be sufficient to cause significant decreases in lift production and increases in drag.

Research has demonstrated that distributed roughness elements having a height of only 1/10,000 of the wing chord can adversely affect performance by increasing stall speeds. This height corresponds to about 0.030 inches on a DC-8-type aircraft - about the roughness of medium to coarse grit sandpaper.

On a wing contaminated with surface roughness, the normal stall progression of a swept wing is altered. The normal nose-down pitching moment in the direction of stall recovery which accompanies a stall is reduced when the wing is contaminated. The effects of the degraded pitching moment characteristics can range from an out-of-trim condition that can result in a different than normal response to control column inputs, to a severe pitch-up as the angle of attack is in increased.

The leading edge portion of the wing is the most sensitive to contamination. Localized ridges, grooves, or narrow bands of roughness near the leading edge of the wing can cause a detrimental effect equivalent to that caused by some lessor degree of roughness elements distributed over the entire surface of the wing. The relative effects of such localized ridges, grooves, or narrow bands of roughness as a function of the location of the roughness expressed in percentage of chord can be seen in Figure 1.16.

Because ice contamination results in a lower than normal stall angle of attack, angle-of-attack-dependent stall warning systems, such as that installed in the DC-8, may not provide warning prior to actual stall.

In 1981 and 1982, the Boeing Commercial Airplane Company conducted wind tunnel, flight, and simulator tests with a Boeing 737 to better understand the effects of wing and horizontal tail contamination on airplane performance and flight characteristics. The results of these studies indicated that contamination significantly reduces wing lift capability, increases stall speeds, and decreases climb performance.

When the wing was contaminated, stall onset flight characteristics occurred within the clean airplane normal manoeuvring envelope, before stall warning system activation.

Flight tests, conducted on a Boeing 737 aircraft with wing surfaces roughened by the application of epoxy potting compound and safety-walk finish with a textured paint roller, demonstrated an 18 per cent loss of maximum lift capacity, which results in a 13-knot increase in stall speed.

Tests in the Engineering Flight Simulator indicated that pilots could encounter stall onset flight characteristics during a normal take-off rotation manoeuvre when the simulator was programmed with contaminated airplane aerodynamic characteristics.

Aircraft without leading edge high-lift devices are particularly sensitive to wing surface roughness. Extension of the leading edge devices on aircraft so equipped will generally recover most of the performance degradation resulting from low levels of roughness.

Unlike the Boeing 737, the Douglas DC-8-63 is not equipped with wing leading edge devices. Douglas Aircraft Co. confirmed that the performance degradation experienced by the DC-8-63 with small amounts of contamination is significantly greater than that encountered by other aircraft types equipped with leading edge devices. Information provided by the Douglas Aircraft Co. indicated that significant reduction in the maximum coefficient of lift and significant increase in the coefficient of drag would be experienced with surface roughness elements of 0.04 inches. (See Figure 1.17.)

According to the Douglas Aircraft Co. data, in an 18-degree flap configuration, the maximum coefficient of lift for the DC-8-63 would be reduced by 25 per cent with wings contaminated by surface roughness elements of 0.04 inches.

In the recent past, considerable research has been conducted into the subject of in-flight ice accretion on airfoils. This research has resulted in the development of several models which can predict the amount of ice that would accrete on a specific airfoil shape under certain conditions. In order to make such predictions, the conditions that must be known include the true airspeed and altitude of the aircraft, static air temperature, liquid water content of the cloud through which the aircraft is flying, and the radius of the water droplets in the cloud.

Several such calculations were performed by CASB investigators and by research officers of the National Research Council (NRC) of Canada. This agency has conducted recent research into icing through experiments undertaken in the high speed icing wind tunnel of the Low Temperature Laboratory. All calculations performed used the airspeeds and altitudes determined from the aircraft FDR recording of the descent into Gander (Figure 1.18.), static temperatures determined from the Atmospheric Environment Service's (AES) rawinsonde released near St. John's Newfoundland approximately two hours after the accident, and liquid water content and droplet size values obtained from AES. The altitude of the top and base of the cloud layer through which the aircraft flew while on approach to Gander was determined from pilot reports made by pilots who either arrived at or departed from Gander both before and after the accident.

One calculation used a method described in the United States Federal Aviation Agency Technical Report: "Engineering Summary of Airframe Icing Technical Data." This calculation resulted in a predicted ice accretion on approach of about 0.25 inch on the outboard portion of the wing.

A second method involved estimating the ice accretion on the wing utilizing a numerical model of rime ice accretion on two dimensional airfoils of arbitrary shape in potential flow. The model was suitable only to provide an estimate of the collision efficiency of the wing and an estimate of the thickness of the accretion which could have occurred at a sufficiently cold temperature. The model predicted that the total collision efficiency of the wing under the assumed conditions would be 0.14, resulting in an ice accretion rate of about 3.8 kilograms per hour per metre of span. The maximum local efficiency was about 0.65 near the leading edge of the wing. Assuming no runback, the local collision efficiency of the wing would have produced an ice thickness of about 0.3 inch.

Both of these methods were limited in value because they assumed a rime ice type accretion on the wing. However, the air temperature during the approach was sufficiently warm that the supercooled drops which struck near the leading edge of the wing would not have frozen immediately on impact, but rather would have run back along the wing surface, freezing at some distance back from their location on inital [sic] impact. The resulting horn-shaped, glaze ice formation would produce a much less streamlined profile than that predicted in the first two calculations.

In an attempt to refine estimates of the probable thickness and shape of wing ice accretion during the approach to land at Gander, a further series of calculations was performed using a method which enables calculation of the thickness of glaze ice formations on a non-rotating cylinder. Since the model was able to simulate icing only upon cylinders, various cylinder diameters were, chosen because they approximated the curvature of the upper and lower surfaces of the DC-8-63 wing profile at various semi-span stations between 26 per cent and 85 per cent.

This series of calculations determined that the most probable estimate of maximum ice accretion expected during the aircraft's approach to land at Gander would vary from 8.7 millimetres (0.34 inches) at 85 per cent semi-span through 6.5 millimetres (0.26 inches) at 53 per cent semi-span to 5.0 millimetres (0.20 inches) at 26 per cent semi-span.

While slight variations between the local collision efficiencies on the cylinder and the DC-8-63 wing would lead to slightly different forms of ice accretion on the wing than on the cylinder, the calculated values were considered to be an accurate approximation of the ice which would have accreted on the wine of the aircraft during the approach to land at Gander.

Interviews with pilots from Arrow Air and other operators indicated that airfoil de-ice is rarely used. For weather conditions similar to those at Gander on the morning of 12 December 1985, it was the consensus of those interviewed that it would be unusual for airfoil de-ice to be used on approach to land.

To comply with the "clean aircraft concept" and minimize the hazards associated with ice contamination, it is common practice in the aviation industry to de-ice an aircraft prior to take-off when conditions warrant. Various techniques of ground de-icing have been developed over the years. Current practice involves the use of Freezing Point Depressant (E;PD) ground de-icing and anti-icing fluids which have the capability to remove ice contamination and provide a protective film to delay further formations of frost, snow or ice.

Ground de-icing facilities using FPD fluids were available in Gander on the morning of 12 December 1985. The crew of MF1285R did not indicate that de-icing was required during their station stop.

Two other aircraft departed Gander in the three hours immediately preceding the accident. A Boeing 737, which departed about 30 minutes before the accident, was de-iced on the ground prior to departure. This aircraft landed at Gander at 0645. During the last two hours of its three-hour station stop at Gander, precipitation in the form of very light freezing drizzle and light snow grains was reported. The individual who de-iced the aircraft reported that ice was present on the leading edge of the aircraft wing prior to de-icing.

A British Aerospace VC-10, which departed about three hours before the accident, was not de-iced prior to take-off. This aircraft spent 50 minutes on the ground at Gander. During that time, no freezing drizzle was reported.

Six ground service personnel attended the aircraft during its technical stop at Gander. All were interviewed by CASB investigators in the days following the accident.

The individual who marshalled the aircraft into position on the ramp and assisted in the placement of the passenger stairs did not observe ice on the aircraft. However, he reported that he did not inspect the aircraft for ice, nor did he at any time go up on the passenger stairs. In response to detailed questioning, he stated that he was not sure if there was. or was not ice on the aircraft.

Two other individuals positioned the passenger stairs at the aircraft, loaded catering supplies and removed trash from the aircraft. Both indicated that they did not observe the wings of the aircraft and could not say whether ice was Present on the wines or not.

A fourth individual serviced the lavatories of the aircraft during the station stop. He reported that he did not specifically look at the aircraft because immediately upon completion of his duties he was required to assist with departure preparations for another aircraft.

One of two refuellers reported that he did not see any obvious need for de-icing but qualified his response to investigators by stating that he could not see the top of the wing and that the top of the wing might have had ice.

The second refueller reported that he did see ice on the edge of the windscreen while he was on the flight deck conversing with the flight engineer. He further reported that the flight engineer had said that the flight had picked up a little ice on descent into Gander. With respect to his observations made while outside the aircraft, this refueller stated that he was unable to see the top of the wing and, thus, could not comment on what ice may have been present.

All six ground service personnel were re-interviewed by CASB investigators later in the investigation. Their recollections of observations made during the aircraft's station stop at Gander were unchanged.

Arrow Air's policy pertaining to flight crew member scheduling is found in the Arrow Air General Operations Manual. The manual states that all Arrow Air flight crew members will be scheduled in accordance with appliable FARs. Arrow Air's Director of Flight Operations testified at the Board's public inquiry. Flight crew scheduling at Arrow Air is his responsibility. He testified that applicable FARs were the primary parameters used to establish individual flight crew schedules. Although no written policy existed with respect to maximum duty day, he considered a duty day of 16 hours to be the upper limit used when developing schedules. He funkier testified that it was normal to allow a two-hour extension to the maximum crew duty day in the event of unanticipated delays and that it was within the discretionary power of a captain to extend the duty day in excess of 18 hours. He also testified that fatigue-inducing factors such as time-zone changes and night departures were not considered in developing crew schedules. The company employed no medical officer or adviser. Pilots were expected to self-monitor for fatigue. Interviews with Arrow Air personnel indicated that, on occasion, pressure was exerted by flight dispatch to complete planned itineraries.

A work book containing a listing of the captain's previous flights was recovered from the wreckage.

One entry pertained to an Arrow Air MFO cargo flight in a DC-8-63, N6161A, to Ras Nasrani,Egypt on 23 February 1983. The entry indicated that the aircraft's main hydraulic system had failed on the night inbound to Ras Nasrani.

The subsequent entry pertained to a night to Cairo after a brief stop at Ras Nasrani. The entry indicated that, in order to get the aircraft out of Ras Nasrani, it was necessary to use the auxiliary hydraulic system to raise the landing gear and flaps. The process was described as slow but successful. Some difficulty was experienced getting the gear doors latched.

The next entry pertained to the departure from Cairo following a fuel stop. It indicated that the aircraft was nursed out of Cairo and that the company diverted the flight to Amsterdam for maintenance. The entry described the approach and landing at Amsterdam and the difficulties encountered (manual reversion of ailerons, cross-wind, and turbulence).

FAR 121.343 requires that large turbine-engine-powered aircraft certificated for flight above 25,000 feet be equipped with one or more approved flight recorders. For aircraft having an original type certificate issued before 30 September 1969, the recorder must record the following information: time, altitude, airspeed, vertical acceleration, and heading.

FAR 121.349 requires that large turbine-engine-powered aircraft be equipped with a CVR. Current regulations do not require that a CVR be functionally tested by flight crews prior to flight.

The Arrow Air, FAA Approved DC-8 Minimum Equipment List permits dispatch of an aircraft with an inoperative FDR and with an inoperative CVR. This is consistent with the FAA Master Minimum Equipment List for the DC-8 and other transport-category aircraft.

The U.S. FAR 121.521 outlines the flight-time limitations for aircraft operated overseas and internationally by a crew of two pilots and one additional airman. It states:

There are no flight crew duty day time limits included in FAR Part 121.

The provisions of FAR Part 121 do not apply when a specific flight does not involve the carriage of persons or property in air commerce for compensation or hire, such as a ferry flight to a maintenance base. Such non-revenue flights operate under the provisions of FAR 91, General Operating and Flight Rules. Under FAR Part 91, there are no flight crew flight-time limitations.

In March 1984, as part of special two-phase National Air Transportation Inspection (NATI), the FAA conducted concentrated inspection and surveillance of air carriers throughout the United States. Findings of the first phase of the inspection were reviewed by regional coordinators who analysed them for trends and potential problem areas. If deficiencies were noted during the phase one inspection of a particular carrier, a second, more detailed inspection of the specific air carrier was immediately initiated. On completion of its phase one inspection of Arrow Air, the FAA conducted a second phase inspection of Arrow Air between 19 March 1984 and 29 March 1984.

The inspection team carried out an in depth review to ensure that Arrow Air operations were conducted in accordance with applicable FARs. The inspection involved the review of records, interviews with company personnel, en route inspections, ramp inspections, and facility inspections.

The inspection followed a period of rapid growth at Arrow Au. Inspectors noted that, in many cases, Arrow Air operating policies and procedures had not kept pace with this growth. Numerous company manuals were found to be out of date, and, in some cases, manuals did not meet the requirements of FARs.

Arrow Air aircrew training records were judged to be unsatisfactory. Many examples of incomplete and unsupported training records were observed. Weak record keeping was a trend found throughout the Arrow Air organization. The inspection team noted that there was no formal maintenance training program in place.

It was determined by the inspection team that Arrow Air operated their fleet of aircraft with many DMIs and that, in some cases, DMIs were carried for months without corrective action.

Inspection of Arrow Air spare parts held in stock at one of their facilities determined that many parts contained serviceable parts tags from unapproved foreign sources. Arrow Air personnel were immediately advised that use of these parts was unacceptable. In response, Arrow Air removed all the parts from stock and shipped them to its Miami base to ensure the parts were properly certified by FAA approved sources.

With respect to FAA surveillance activities, the inspection team noted several cases where routine surveillance had identified discrepancies which appeared to be violations of FARs, and where the inspection results were recorded as satisfactory.

There were also instances where no follow-up action had taken place after unsatisfactory surveillance observations had been made. There were other instances where the inspection team determined that the carrier itself had been asked to investigate alleged violations of FARs.

As a result of its observations, the inspection team made a number of recommendations. Many pertained to increased surveillance on the part of the local FAA office and the assigned principal inspectors. They recommended that operations and maintenance units increase their surveillance and inspections of the carrier; that all discrepancies or unsatisfactory findings noted be followed up to ensure that corrective action was taken; that, where deemed appropriate, violation action was taken; and that, in general, a more firm stance be taken with respect to company activities and practices found to be inappropriate or contrary to regulations.

Specific areas identified as requiring increased surveillance activity and follow-up were company manuals, training, DMIs, and the use of parts from unapproved foreign sources.

The FAA conducts ongoing surveillance of air carriers to ensure compliance with FARs and approved FAA procedures. The responsibility for this ongoing surveillance is primarily that of the assigned POI and principal maintenance inspector (PMI).

The FAA inspectors who were the assigned POI and PMI at the time of the accident testified at the Board's public inquiry into the accident.

The assigned POI had assumed his responsibilities at Arrow Air in June 1985, six months before the accident. He testified that about 75 per cent of his duties related to activities at Arrow Air. Of that time, only 30 per cent was devoted to inspection and surveillance, the remainder to providing technical advice to the carrier. Although an assistant operations inspector position existed, it had been vacant since the POI was assigned to Arrow Air.

In the opinion of the POI, Arrow Air's operations met the requirements of FARs and approved FAA procedures. He further testified that he believed the manpower resources available to him for surveillance of Arrow Air were inadequate.

The assigned PMI had assumed his responsibilities at Arrow Air in April 1985, eight months before the accident. He testified that it was his responsibility to ensure that Arrow Air maintenance and inspection programs were in accordance with FAA standards. Surveillance activities would include spot inspections, en route inspections, facility inspections, review of various manuals and aircraft technical logs as well as aircraft inspection. He further testified that, in addition to his duties at Arrow Air, he was also the PMI for two other air carriers and five repair stations. He estimated that about 40 per cent of his time was devoted to surveillance of Arrow Air. To assist him in his responsibilities were two other FAA inspectors.

During his eight months of surveillance at Arrow Air, the PMI did not identify any deficiencies in Arrow Air's maintenance and inspection programs. He considered Arrow Air's operations to be in accordance with required standards. He further indicated that, in his role as PMI at Arrow Air, he did not in any way use the report of the 1984 NATI of Arrow Air.

In January 1986, one month after the accident, the U.S. Secretary of Transportation directed the FAA to conduct in depth inspections of airlines operating under military charter. Arrow Air was subject to such an inspection between 21 January 1986 and 21 February 1986.

The inspection team made numerous observations which were, in the opinion of inspectors, instances of non-compliance with FARs or accepted FAA procedures. In several cases, findings were similar to those determined in the 1984 NATI of Arrow Air. Inadequacies and examples of non-compliance with FARs were noted in almost all areas of Arrow Air's operations. Specific observations included out-dated manuals, procedures not in accordance with FARs, unsatisfactory training files, non-compliance with established Arrow Air procedures, use of aircraft parts and components from unapproved foreign sources, and non-compliance with FAR maximum flight-time limitations and minimum crew-rest provisions. Although no overall conclusions were drawn as a result of this inspection, the FAA inspector in charge testified at the Board's public inquiry that, in some areas, Arrow Air did not meet the minimum standards required by the FAA. He further testified that he considered some of the observations made to be significant and that, in some cases. the safety of operations was questionable.

Subsequent to the public inquiry, the FAA informed the Board that, after an in depth review by the FAA's Miami Flight Standards District Office, many of the inspection team's findings were found to be invalid for a variety of reasons. Specifically, many of the findings were considered to be of a minor nature, and, of the 19 findings considered by the FAA to be major, that is, worthy of formal enforcement proceedings, further investigation determined that eight were not violations, and they were subsequently dismissed without further action. Ten of the major findings were determined to be violations which resulted in assessment of a civil penalty or issuance of a warning/correction letter.

The FAA further indicated that, when compared against in depth inspections carried out at other carriers, the magnitude of Arrow Air noncompliance was no worse than "average", indicative of violations in limited areas of their operation.

The FAA also asserted that, in the months preceding the accident, their surveillance and follow- up of Arrow Air was executed to a greater degree, both in quality and quantity, than ever before in the company's history.

The CASB conducted a seven-day public inquiry into this accident in Hull, Quebec, beginning 08 April 1986 (See List of Witnesses - Appendix D). Participants in the inquiry were the CASB technical panel; Douglas Aircraft Co.; Pratt & Whitney United Technologies; the Flight Crew Next of Kin; Arrow Air Inc.; Multinational Force and Observers; Department of Transport, Canada; United States Federal Aviation Administration; United States National Transportation Safety Board; United States Army; and the Department of Justice, Province of Newfoundland.