• Print

The Fall of EgyptAir 990



On July 17, 1996, TWA 800 took off from JFK International Airport; twelve and a half minutes later, it fell into the ocean south of Long Island. On September 2, 1998, Swissair 111 took off from JFK Airport; fourteen minutes later it lost radio contact with air controllers; it continued flying north, eventually regained radio power, reported smoke in the cockpit as it neared Nova Scotia, then fell into the Atlantic Ocean. On October 31, 1999, EgyptAir 990 took off from JFK Airport; it flew east for thirty-one minutes, then suddenly dove into the ocean east of Long Island, south of Nantucket. The 676 people on board the three planes perished. No other large passenger plane taking off in the United States crashed during this three years and three months period.

The uniformity of the region in which the accidents occurred suggests that the region itself—the environment external to the plane—should in each case be included among the causes to be investigated. Substantial studies by the Joint Spectrum Center and by the National Aeronautics and Space Administration (NASA) have gone a long way toward reconstructing the electromagnetic environment of the first accident; but the work on behalf of TWA 800 is still incomplete, and in the cases of Swissair 111 and EgyptAir 990 it has barely begun.

The need for an external reconstruction is suggested not only by the uniformity of the region in which the accidents have occurred, but by the absence of any definitive internal cause: TWA 800 is believed to have fallen as a result of a central fuel tank explosion, but the ignition source of that explosion has not been found. (At its hearing on the accident this August, the National Transportation Safety Board divided potential ignition sources into “likely” and “unlikely” and designated an electrical “short circuit” as the least unlikely; but investigators repeatedly acknowledged that they have not found definitive evidence of that short circuit or even a probable location.)

Swissair 111’s fall is believed by both the Transportation Safety Board of Canada and Boeing to be electrical in origin, but no one system on the plane, or section of wiring, or even segment of the flight has as yet been ruled out as a cause. EgyptAir 990, according to the United States National Transportation Safety Board (NTSB), suffered no mechanical malfunction that can account for its sudden dive toward the ocean: the absence of a mechanical cause has been repeatedly cited by the Safety Board and by the press as evidence that one of the pilots must have intentionally chosen to kill himself and all his fellow travelers (a point that I will return to).

The first part of the present article1 looked at the features shared by TWA 800 and Swissair 111. Each appears to have suffered an electrical catastrophe, but the cause of each crash has so far proven elusive. Each took off from JFK; each traveled east along Long Island on the “Bette” route (which lets planes avoid a region of ocean space, called W-105 and W-106, when it is reserved for military exercises); each took off on a Wednesday at 8:19 PM; each experienced its first difficulty twelve to fourteen minutes into the flight (a fatal difficulty for TWA 800, a radio difficulty of not-yet-understood gravity in the case of Swissair 1112 ); each flew during a week in which military exercises were scheduled; each flew on a night when submarines and Navy P3s were operating off the coast.

EgyptAir 990, like TWA 800 and Swissair 111, took off from JFK and crashed during the opening segment of its transatlantic journey. Like them, it suffered from a series of events that—as we will soon see—are compatible with an electrical accident.3 But it should be stated from the outset that EgyptAir 990 does not share all the external features that overlap in the accidents of the other two planes. It took off in the middle of the night, at 1:19 AM on a Sunday, not on a Wednesday at 8:19 PM. It did not travel, as the other two did, on the Bette route, which hugs the Long Island coast in order to skirt the northern edge of military exercise zones W-105 and W-106. (These are zones that are identified on flight maps as follows: “Warning: National Defense Operating Areas, Operations hazardous to the flight of aircraft conducted within these areas.”)


EgyptAir 990, as the map shows, instead flew directly into and through military exercise zone W-106, and then crossed into exercise zone W-105, where it flew approximately 120 miles before diving irrevocably toward the sea. Its crash site lies deep within military exercise zone W-105.

When EgyptAir 990 left JFK Airport, it was initially assigned a route that would have taken it down and around the military exercise zone W-105, a route described by the air controller as “Shipp Linnd Lacks Dovey.”4 The four names indicate a sequence of ocean intersection points; and as the map shows, had EgyptAir 990 flown to Shipp, then Linnd, then Lacks, then Dovey, it would never have entered military zone W-105.5 But it is normal practice, once a plane is in flight, for air controllers along its path to give it a more direct route, if the shorter route is clear of traffic or, in the case of warning zones, if the zone has been released by the military to the FAA. By 1:35 AM EgyptAir 990 had passed through two New York radar sectors, Kennedy and Manta, and had arrived at the third, Atlantic, which monitors the opening section of a transatlantic flight. A few seconds before 1:36 AM the Atlantic controller cleared EgyptAir 990 to fly directly to ocean point Dovey, without first passing Linnd and Lacks:

Atlantic Controller: EgyptAir 990, climb and maintain flight level three three zero [33,000]. Cleared direct Dovey.

EgyptAir 990: Three three zero, direct Dovey, EgyptAir 990.

On the map, this is the moment where the plane, heading south toward Linnd, now abruptly turns east and begins to fly across the W-105 zone.

To many people—certainly to this writer—it seems common sense that when a civilian plane crashes, the number and nature of all electromagnetic transmitters in the area at the time of the crash should be quickly identified. It seems common sense that this should be an automatic practice, just as many other forms of evidence-gathering at once go into effect when a plane falls. (In the first installment of this article, the former director of the Defense Department’s Joint Spectrum Center recommended that this be done.) When a civilian plane crashes in or near a military exercise zone, there is evidently all the more reason to ensure that such a comprehensive identification of transmitters be made.

No special trace of a maintenance problem, or a fuel problem, or a baggage compartment problem is needed to spur the NTSB to gather maintenance records, fuel samples, and the cargo manifest; the gathering of this material automatically begins the very day a large passenger plane falls. It should similarly not require special evidence or argument to ascertain the location of transmitters in the external environment.

The fact that EgyptAir 990 was permitted to fly through W-106 and W-105 should mean that no military exercises were occurring in those areas; and the NTSB has made available to the public the FAA daily logs which confirm that the military exercise zones were open to civilian flights during the hour (or what in the case of W-105 turned out to be a quarter of an hour) that EgyptAir 990 flew there. The NTSB has also made available to the public the record from Fleet Area Control and Surveillance Facility in Virginia Capes (this facility is often abbreviated FACFACS VaCapes, but it refers to itself throughout its own internal documents and its communications with the FAA as “GIANT KILLER”).

The records of this facility show only one exercise scheduled in W-105 on October 31, much later in the day than the 1:19 AM flight. But the absence of scheduled military exercises does not guarantee that no military craft were in the corridors around the warning zones or inside the warning zones.6 Military planes flying inside W-105 will not show up on FAA air controller tapes because at the point of entering the region they “go operational,”meaning they cease to be in communication with controllers. The possible dangers within the warning zones and corridors would arise not from munitions but from high-powered radar or other electromagnetic signals from fixed ground transmitters or from transmitters on craft passing through the area.

We already know (from news reports during the week immediately following the accident) that powerful Air Force radar antennas were tracking EgyptAir 990. Because these transmitters are designed to spot cruise missiles, they have capacities not shared by any civilian radars. A civilian radar can read the altitude of a plane only if that plane has a functioning transponder, the device that identifies the location of a plane. The Air Force antennas, in contrast, are designed to identify the altitude of a flying object, even when the flying object itself has no transponder, either because it wants to disguise its altitude (as in the case of a cruise missile or again in the case of a Navy or Air Force plane carrying out an exercise) or because it has lost the use of its transponder (as occurred in the case of EgyptAir 990, as well as TWA 800 and Swissair 111, before they crashed).

EgyptAir 990’s dive into the sea took place in three phases: a plunge from 33,000 feet down to 16,000 feet, a sudden rise back up to 24,000 feet, then a final plunge to the bottom of the sea. EgyptAir 990 lost its transponder toward the bottom of its first dive. It is only because the aircraft was an object of scrutiny on the part of the Air Force antennas at Riverhead, Long Island, and Truro, Cape Cod, that we know about the last two segments of flight—the sudden rise and the final fall.7 What other transmitters were observing EgyptAir 990, and what other transmitters were communicating with one another in the vicinity of EgyptAir 990, are questions that need to be answered.8

In recent years the military has increasingly attended to “littoral,” or coastline, warfare, as opposed to warfare taking place on the open seas, the focus of military practice in an earlier period. One of the constant tasks of powerful military radars—those carried on Aegis cruisers, for example9—is perfecting their ability to track a small flying object and, crucially, to hold the image of that flying object distinct from “the clutter” of civilian coastline life. Inevitably, then, civilian planes are being monitored by such transmitters and antennas: they are part of the coastline clutter from which real targets must be differentiated. Under what circumstances, and during which periods, civilian flights become a special target of observation is a subject not widely discussed in the United States, and one that ought to be.10

It is not hard to think of reasons why EgyptAir 990 might have become an ordinary—or even an extra-ordinary—target of intense observation. EgyptAir 990 was, first of all, a foreign carrier from a country that is not always regarded as a close United States ally. Second, it was carrying, in addition to its civilian passengers, thirty-three Egyptian military officers, including one brigadier general.11 During their visit to the United States, the thirty-three officers were, according to the Defense Department spokesman, Kenneth Bacon, subdivided into six distinct groups: the members of one group were here to study communications; a second group was testing recently purchased helicopters; the third group was here to receive “training on high frequency” equipment; the fourth group was learning about telecommunications; the fifth group was studying the repair of Chaparral missiles; the members of the sixth group were carrying out individual projects.12 Whatever their separate missions, they had all convened to take this particular flight for their return to Cairo.

A third possibility is that EgyptAir 990’s passage through W-105 might have come as a surprise to the military and might therefore have provoked special attention. Any one of the three possibilities could have operated in isolation,13 or in combination14 with one another. (And, of course, none of the three may have taken place: the plane may have been tracked by the Air Force radars simply because all civilian planes are at all times being similarly tracked.)

But what of the third possibility I have mentioned: What reason is there to think that EgyptAir 990 could have surprised military observers in the region (and therefore become a closely scrutinized object)? Just this: the flight surprised at least one civilian air controller that night. The automatic computer reports that would ordinarily have made EgyptAir 990’s passage from one sector to the next wholly unsurprising for civilian controllers were not functioning flawlessly; perhaps the communications between civilian and military controllers were also flawed.

At the point where the plane was being handed off from the New York controller at Kennedy to the New York controller at Manta, the Manta controller expresses dismay and impatience that the flight has not been entered into the computer:

Kennedy Controller: Manta, Kennedy manual handoff, EgyptAir 990.

Manta Controller: Doesn’t anybody know over at the tower that they’ve got to put these flight plans back in?15

The NTSB provides among its documents a description of the fact that, as part of a normal procedure, the computers had been shut down for a routine repair. It is true that the controllers don’t seem to find their situation abnormal (“Let me see if there’s anything in here. Of course not. And I don’t have all his routing either. Oh, that’s wonderful”). But neither do they find it acceptable (“It’s just disgusting”). There is no question that the handoff from the Kennedy to the Manta sector is safely executed: the Manta controller does not have the route in the computer, but the Kennedy controller recites it orally (“Shipp Linnd Lacks Dovey”); the Manta controller does not have the pilot’s requested altitude, but the Kennedy controller provides it (Manta: “Do you know what he wants as a final?” Kennedy: “I’ve got thirty-three thousand”).

This first phase of the air controller’s complaint lasts slightly less than one minute and should probably be seen against the backdrop of complaints in every other line of work. But the point is not whether anything wrong happens here; the point is that if the flight isn’t in the Manta controller’s computer, will it—at the point ten minutes later where the flight is directed by the Atlantic controller to veer east into W-105—be in GIANT KILLER’s computer? A July 15, 1999, agreement between the East Coast FAA air controllers (“hereinafter designated Center”) and the facility (“hereinafter designated GIANT KILLER”) specifies that even when the warning zones are released to the FAA, the civilian planes “penetrating” the air space have to be announced to the military by explicit procedures, such as computer notification.

The conversation between the Manta and Kennedy controllers comes to a halt because the Manta controller has to exchange information with two planes, first EgyptAir 990, then Costa Rica Airlines 661, called a “Lacsa.” But five minutes later the Manta controller is still upset and returns to the subject, asking the Kennedy controller: “You got any more surprises after this Lacsa coming off, or is he the last one?” An important clarification now follows. What has upset the Manta controller is not that all the flight reports have failed to be put in the computer. The problem is that they have been entered there inconsistently:

Manta Controller: …Nobody typed in the EgyptAir but they did type in the Lacsa.

The Kennedy controller tells the Manta controller that he has no “tickets” (advance information) about planes that he will be forwarding to the Manta sector, but he then goes on to note that the lack of a ticket doesn’t mean there won’t be any more planes: “Just because you don’t have a ticket on anybody doesn’t mean there’s nobody else.”

Manta Controller: …So, we’re both gonna be in the dark.

Kennedy Controller: Well, there ya go.

The element of surprise audible in the Manta controller’s statements to the Kennedy controller (“You got any more surprises…?”; “Nobody typed in the EgyptAir but they did type in the Lacsa”; “We’re both gonna be in the dark”) is registered in at least one other radar sector as well.16

Small tremors of unease about the plane’s route occur at various places in the air traffic control record. While the plane is still on the ground back at JFK, the Kennedy Tower ground controller expresses concern. The plane is two hours late taking off; does the filed flight plan still hold?17 He consults the controller in charge of flight data (“Is that flight plan still good?”) and is assured that it is (“Yeah. That’s still good”). Should the plane be given a new transponder code? He is assured: “The code that he has is good.” The time is 12:56. Twenty-three minutes later (at 1:19 AM), when the Kennedy Tower local ground controller transfers control to the New York traffic controller, he recalls his original worry: “I’ve got an EgyptAir 990 going over Shipp…. It’s on an older strip; Center says it’s still good.”18

Later, when the plane is being placed in the Manta controller’s hands, the Manta controller expresses lack of confidence about using the computer to pick a flight path: “OK. Let me start a track…pick. This new equipment—I don’t even know how to do this stuff. Enter. There he is. OK. Interim two three oh [altitude 23,000 feet].”19 Again, these passages are cited not to suggest any transgression or lapse of attention on the part of the air traffic controllers, who seem alert and concerned, but to suggest their shared unease about sending EgyptAir out into a night on which everyone seems to be somewhat “in the dark.”

Was EgyptAir 990 the only civilian plane flying through the W-105 region, as appears to be the case in the two-hour20 period covered by the air controller transcripts? (The question is worth asking because a solo plane entering W-105 might startle military onlookers in a way that a steady stream of civilian planes would not.) If EgyptAir 990 was indeed the only civilian plane in the warning zone, there seems to exist a simple and straightforward explanation: it is late at night; few planes are flying; and most of those that are flying are not embarking on a transatlantic flight. Costa Rica Airlines 661, heading for home, flies straight down the coastline; Lufthansa 8202, a cargo plane which had stopped at JFK in its travel from Frankfurt to Atlanta, now flies south-west over Robbinsville, New Jersey, and continues south. There is, however, one other flight setting out on a transatlantic journey, the cargo plane El Al 2812. It prepares for takeoff almost simultaneously with EgyptAir 990. The controller begins to instruct El Al 2812, traveling to Tel Aviv by way of Frankfurt, to exit from JFK Airport by the Kennedy seven departure lane (the same gateway EgyptAir 990 flies out of). But the controller then instructs the pilot to follow a different path:

Kennedy Tower Controller: El Al 2812, good morning, you’re cleared to Frankfurt via the Kennedy departure seven—Actual[ly], it’s a Bette two departure and Nantucket transition….

El Al 2812: Roger, El Al cleared to Frankfurt via Bette two Nantucket transition gateway….

The air route from JFK to the Bette intersection, and from Bette to Nantucket, is the path taken by transatlantic flights when W-106 and W-105 are in use by the military. (It is the path both TWA 800 and Swissair 111 traveled.) The fact that this route is used when military exercises are underway does not mean that the path cannot be used when the W-106 and W-105 regions are open to civilian planes. But the fact that the cargo plane’s flight route was along the Bette path and EgyptAir 990’s was not (like the fact that Costa Rica Airlines’ flight was typed into the computer and EgyptAir 990’s was not) calls attention to the isolation of EgyptAir 990 on its middle-of-the-night voyage home.21

Given the higher level of danger that exists in a military warning zone than in open seas, prudence and ordinary investigative precision would seem to mandate a reconstruction of that environment. Details from the route have been given here not because they show something wrong,22 but because they are a reminder that the world external to the plane is at least as complicated as the one inside the plane that has already for many months now undergone intense review.


The EgyptAir 990 crash, according to repeated NTSB announcements, involved no mechanical mishap, or at least no mechanical mishap that has so far been recognized on its sophisticated data recorder.23 Electromagnetic interference should be scrutinized as a possible cause, even in plane crashes where a mechanical explanation seems to exist, since the mechanical event itself may have been triggered by a prior electrical event24 ; but the absence of any mechanical explanation (as so far appears true of EgyptAir 990) should increase the rigor with which data concerning radio and radar transmitters in the area is gathered and studied.

It is striking that the tiny handful of stark facts known about the fall of Egypt Air 990 are themselves consistent with electromagnetic interference. Here are the four facts we know. One: the plane’s autopilot disconnected. Two: the plane went into a sudden steep dive. Three: the elevators on the plane’s tail, the movable surfaces on the horizontal portion of the tail that control the pitch of the plane (by determining whether the plane’s nose points up, down, or level), acted anomalously, moving independently rather than in coordination with each other. Four: the two engines turned off (shortly before the transponder, and presumably all other electrical systems, lost power).

These four features of the EgyptAir 990 catastrophe are all prominent in the most important literature on electromagnetic interference—the 1988 Air Force study of military craft and the 1994 NASA study of civilian craft by Martin Shooman.25 I will consider them in turn, beginning with the disconnecting of the autopilot.

The 1994 NASA study contains an appendix in which individual pilots summarize incidents caused by electromagnetic interference. Here are excerpts:

Autopilot malfunctions commuter airliner…

Manual A/P [Autopilot] disconnect…

Unmotivated A/P [Autopilot] disconnect, business transport…. In stable cruise, sudden A/P [Autopilot] disconnect resulting in -5 pitch down…

Helicopter reported interference to Auto Pilot and rotor controls in vicinity of Navy Ship with high power. Problem associated with cable pick up of RF [Radio Frequency]. Same problem identified several times.26

How does an autopilot disconnect in situations of electromagnetic interference? There are two ways. The auto-pilot (as in some of the examples above and, again, as in the late phase of the Swissair 111 accident) may automatically disconnect, either because it receives a stray electrical command which shuts it off or because it becomes baffled by the random information it is receiving and so shuts down. A second possibility is that the human pilot sees that an order he or she did not authorize is starting to be carried out on one of the plane’s control surfaces (such as its rudder, elevators, ailerons, or wing flaps). When this happens, it is appropriate—indeed necessary—for the pilot at once to disconnect the autopilot. (The second passage above describes pilot intervention.) It is perfectly plausible that the EgyptAir 990 autopilot may have disconnected on its own27 ; it is equally plausible that the pilot, seeing something amiss, quickly and appropriately disconnected the autopilot.

The NASA researcher Martin Shooman points out that the disconnecting of the autopilot, whether executed by the autopilot itself or by the live pilot, becomes necessary to the plane’s survival (since the autopilot is starting to put the plane on a disastrous course). But, at the same time, that act of disconnecting is itself an especially hazardous event. Up to that moment, the autopilot has been receiving false instructions (such as an adjustment to the elevators, or a deflection of the rudder); but if the errant instruction is incompatible with safe flying, the autopilot will simultaneously start compensating for the false command with balancing counteractions. When suddenly the autopilot is turned off, the human pilots are abruptly confronted with a set of events that may be confusing, and there may not be time either to put in place or to undo the compensatory settings on other control surfaces.28

Eight seconds after EgyptAir 990’s autopilot disconnected, the plane entered a sudden steep dive. Especially telling on this matter is the previous work carried out by the Air Force on electromagnetic transmissions. Although the 1988 Air Force study, and the three-year-long Pentagon study it prompted, have still not been released to the public, the study’s major findings were summarized by Colonel Charles Quisenberry, one of the study’s authors: he stated that electromagnetic interference can send a plane or helicopter into an uncommanded roll or dive.29 Electromagnetic interference can take scores of different forms. But Colonel Quisenberry did not list scores of outcomes. Because he was giving a summary, he listed only the three most important, one of which was a sudden, uncommanded dive. Sudden dives are also registered in the 1994 NASA study of civilian craft: one commercial airliner dropped 30,000 feet; a second commercial airliner dropped 20,000 feet.30

The prominence of inexplicable dives in accidents involving electromagnetic interference makes it seem odd that the NTSB has so far not named it among possible causes in need of investigation in the case of EgyptAir 990, especially given the fact that this same geographical region had been implicated in two earlier plane fatalities, thereby increasing the possibility that the cause of all three involved the external environment.

The third key event (the position of the aircraft’s elevators) is closely related to the second. Elevators are among the surfaces whose control can be interfered with by electromagnetic transmission. When Colonel Quisenberry reported that planes can enter a sudden dive, he meant that some control surfaces on the plane (elevators, rudder, wing flaps) can be altered in such a way that the plane goes out of control and begins to fall.31

The fourth event, the sudden shutting down of the engines, is also a prominent part of the literature on electromagnetic interference. If we return to Colonel Quisenberry’s stark summary of the Air Force study, we find included in his short list of key events the sudden interruption of the fuel flow to the engines.32 This can be rephrased as “sudden turning off of the engines” since when a pilot (or a stray electrical signal) turns off the engine, what that command does is interrupt the fuel flow. Reports of the sudden turning off of engines also appear in the 1994 NASA study: a blimp suddenly had a double-engine failure at the moment when it was passing directly over a powerful Voice of America antenna in North Carolina.33 In another case, one engine of a plane ceased at the same time that the autopilot disconnected when in an area of strong transmissions.34

Because human beings and stray electrical signals can each introduce “commands” into the control surfaces of a plane, it is extremely difficult to sort them out, and at the same time crucial to do so.35 Decoupling the two is difficult enough in the early stages of the catastrophe; it is even more difficult once the catastrophe is fully underway. Once EgyptAir 990 had entered its steep dive, it may be that all subsequent command actions were initiated by the pilots, desperately trying to regain control of the plane and making the best decisions they could: those actions may have been a highly competent effort to save the aircraft or they may have been, however admirable, less competent. It is possible that all four of the key events—automatic pilot shutdown, steep dive, split-elevator anomaly, engine shutdown—were caused by electromagnetic interference or that only some of the four were caused by electromagnetic interference, the rest by pilot intervention and attempted rescue.36

Electromagnetic interference affects different planes in different and sometimes quirky ways because it interacts with a particular plane’s greatest points of electrical vulnerability: if a plane has chafed wiring (as in the case of TWA 800), it may produce a problem there; if a plane has an over-ambitious entertainment system (as has sometimes been claimed about Swissair 111, though the Canadian Safety Board has repeatedly stated that it has, for them, no more prominence in the accident than any of the plane’s other electrical system), the problem could first appear there. If it was a highly digitized plane (such as EgyptAir 990’s plane type, the 767), it could interact by introducing a false command into the autopilot, ignition, stabilizer, or other control surfaces.

The National Transportation Safety Board is usually scrupulous about not claiming to know a cause before having the evidence. It can sometimes be faulted for omissions, such as not reconstructing the external environment, or not bringing the same standard of inquiry to bear on military actions that it brings to bear on the actions of wire manufacturers or maintenance personnel or civilian passengers. But it would be hard to describe it as ever committing a faulty act. Its stamina in moving through four years of investigating TWA 800 without ever prematurely claiming to have found a definitive cause is remarkable.

It is therefore almost incomprehensible how this same Safety Board has come so close (at least as its statements are reflected in the media) to accusing the copilot of EgyptAir 990, Gamil al-Batouti, of willfully murdering 216 fellow travelers. The Safety Board officials acknowledged before Congress and before the public that they so far have insufficient evidence to hold Mr. Batouti responsible for the catastrophe; but to many onlookers (including this author) what they call insufficient evidence looks instead like no evidence at all.37 It certainly does not appear to be the case that only a pilot’s willed action could bring about the disastrous events (autopilot disconnect, dive, split-elevator anomaly, engine shutdown) that took place that night.

By gathering and providing to the public transcripts of the air controllers’ tapes, FAA logs, communications from the Virginia Capes Surveillance Facility, and many other materials, the NTSB has already begun to carry out a reconstruction of what might have happened. It is to be hoped that this is the first step in what will eventually become a comprehensive review of the external electromagnetic environment. Such a review has still to be carried out for the three planes that have, since July 1996, crashed after leaving JFK Airport.

—This is the second part of a two-part article.

…We have a variety of radar around the country that is used to defend our territorial integrity, to monitor air traffic. Obviously, there’s a huge amount of air traffic coming in and we monitor it all the time…and special monitoring—

I suppose they do exercises and they do a variety of other things, trying to improve their capabilities.

  1. 1

    Published in the September 21, 2000, issue of The New York Review. The complete text with full citations from official documents appears on the New York Review website, www.nybooks.com. The same website will soon carry an appendix to the two-part article describing the possible sources of electromagnetic interference with TWA 800 that remain to be investigated. 

  2. 2

    Although the Transportation Safety Board of Canada has not singled out the radio blackout as a focus more important than any other part of the flight, it is certainly among the objects of their concern. On March 9, 1999, the Safety Board sent an “Interim Air Safety Recommendations” letter to David Michael Collenette, Canada’s minister of transport, urging that the duration of aircraft cockpit voice recorders be extended from their current length of one half-hour to a much longer period. 

  3. 3

    The TWA 800 and Swissair 111 accidents, however, appear to be “unmistakably” electrical; in the case of the EgyptAir 990 accident there is persuasive, but not unmistakable, evidence that the accident could be electrical. 

  4. 4

    The words were used by the New York controller at the Kennedy Radar Sector when speaking to the New York controller at the Manta Sector, toward which EgyptAir 990 was flying (1:25 AM on transcript of FAA tape of New York TRACON, Kennedy Radar Sector, October 31, 1999, 1:16 AM to 1:31 AM, Addendum to Air Traffic Control Group Factual Report, NTSB, William English Chair, p. 3; and again on the transcript for the Manta Radar Sector, covering 1:19 to 1:39, p. 2). All transcripts, FAA logs, and military logs quoted in this section of the article are materials made available to the public by the NTSB on its accident website, www.ntsb.gov/events/EA990 (released on August 11, 2000). 

  5. 5

    Needless to say, EgyptAir 990 may have crashed even had it followed a different route. But this article is concerned to examine the possibility (a possibility that has so far not received public attention) that the specific environment the plane was in—hence its route—may have had a crucial part in the accident. 

  6. 6

    In fact, the single plane other than EgyptAir 990 with which the Atlantic controller is in contact during the period before the crash is a Navy plane with the call sign “Arise 57,” which was flying in a corridor between warning zones. (It is at the moment that Arise 57 checks in that the control-ler, who had briefly stepped away from the radar screen, returns and realizes that EgyptAir 990 has disappeared from radar and radio.) Arise 57 appears to be flying at a great distance—approximately 150 miles—from EgyptAir 990: the transcript indicates a flight path from ocean intersection point Champ to ocean fix Sea Isle (these two points are south of the region shown on the map). 

  7. 7

    The role of Air Force transmitters in capturing the image of EgyptAir 990 was widely reported in the United States media, as was the work of the 84th Radar Evaluation Squadron at Hill Air Force Base in Utah, which put together the images captured from different Air Force radar locations. The specification of Riverhead, Long Island, and Truro, Cape Cod, was less widely reported: it appeared in Matthew L. Wald’s November 4 and 5, 1999, New York Times stories ( the first of which was reprinted in newspapers in many states) as well as in November 4 reporting by CNN’s Carl Rochelle and Lou Waters. 

  8. 8

    At his November 4, 1999, press conference, Defense Department spokesman Kenneth Bacon was several times asked by a persistent reporter what activities the Air Force radars and their operators normally carry out (when not assisting in the analysis of a civilian plane crash). According to Mr. Bacon, the work of the Air Force radars includes both continuous blanket monitoring— 

  9. 9

    For a description of such transmitters, see my article “The Fall of TWA 800: The Possibility of Electromagnetic Interference,” The New York Review, April 9, 1998. 

  10. 10

    Should the civilian population become involved in military exercises without their knowledge or authorization? While it would not be possible for citizens to give their authorization on a flight-by-flight basis, certainly the general issue could be discussed and debated. The only public discussion so far came in the spring of 1997, when military planes followed so closely behind civilian planes in tracking exercises that the affected commercial pilots threatened to strike if the practice continued. 

  11. 11

    I have specified one brigadier general, even though the number varies, depending on the news source, from one to three. According to Jane’s Defence Weekly, the group of thirty-three officers included “two Egyptian Air Force brigadier generals and two army major generals” (Vol. 32, No. 19, November 10, 1999). 

  12. 12

    Defense Department, Pentagon News Briefing, November 4, 1999. 

  13. 13

    Taking the second as an example: United States military observers may have been aware of the plane and its special passengers long before the plane left the ground and all along its route. (Kenneth Bacon has stated that the presence of thirty-three officers is not unusual: one thousand Egyptian officers come to the United States each year.) 

  14. 14

    For example, if United States military observers were notified of the plane’s approach only very shortly before it arrived, they might give it special scrutiny. Once they began to scrutinize it, they would realize it was an Egyptian carrier. They might give it even closer scrutiny while checking the background of the flight, and discovering that thirty-three military officers were on board. 

  15. 15

    FAA transcript of Manta Sector controller. (Here and throughout I have eliminated all “uh”s and “ah”s between words registered in the transcripts; and I have added punctuation.) 

  16. 16

    The Atlantic controller, for example, has no record in her computer of the Navy plane Arise 57; so the Kraft controller faxes a record to her (“one invalid field record…on its way”). (Minute 1:32, FAA transcript for Atlantic Sector.) 

  17. 17

    The clocks are changing from Eastern Daylight Savings on this night to Eastern Standard, and once the clock is set back, the flight delayed by two hours seems delayed by only one hour. 

  18. 18

    FAA transcript for New York TRACON sector, 1:16 AM to 1:31 AM. 

  19. 19

    Minute 1:25, FAA transcript for Manta sector. 

  20. 20

    The period from 12:48 AM to 2:52 AM is covered by the five transcripts. Each only covers a segment of that period, and it is possible that the two-hour record for each of the five would show other planes traveling through the W-105 and W-106 regions. 

  21. 21

    The Atlantic controller, as mentioned previously, enlists the help of many different civilian and military resources once she realizes the plane has disappeared from radar. Among them are Lufthansa 499 (a nonstop flight from Mexico City to Frankfurt) and Air France 439 (a nonstop flight from Mexico City to Paris), the first of which tries to contact EgyptAir 990 on the radio and the second of which is actually conscripted into digressing from its own route to fly over a patch of the ocean to see if any sign of the plane can be seen. But by the time these two planes are called on for help, EgyptAir 990 has already been under the sea for a long time (thirteen minutes in Lufthansa’s case; forty-five minutes in the case of Air France). Though various searchers are looking over Dovey, northeast of Dovey, and even as far off course as W-102 (see map), by the time the controller is in contact with Air France she appears to have inferred what will eventually turn out to be the actual accident site, for she requests that he search for the plane “sixty miles south of Nantucket,” the description now widely used for describing the location of the fall. (Minute 2:32, transcript of Atlantic sector.) 

  22. 22

    The air controller tapes from the Swissair 111 flight, in contrast, show something unequivocally wrong, even if we cannot yet judge the relation between the plane’s thirteen-minute radio blackout and its eventual fall. 

  23. 23

    Although the initial causes of the TWA 800 and Swissair 111 accidents have not been found, there are recognizable physical problems along the way: an explosion in the central fuel tank in the case of TWA 800, “arcing”—a hot and persistent form of electrical sparking—in twenty wires located near the cockpit in the case of Swissair 111. Nothing equivalent has been found in the case of EgyptAir 990, except the four events involved in the final dive, looked at below. 

  24. 24

    Two examples involved a Boeing 767-300, the type of plane involved in the EgyptAir 990 crash. In 1993, a 767 attempting to land at Frankfurt suffered a sudden rudder deflection of 16 degrees which sent it off the runway. The event was judged to have been instigated by electromagnetic interference (“Government Officials Investigated United 767 Rudder Incident,” Aerospace Daily, September 21, 1993, Vol. 167, No. 57, p. 473). 

  25. 25

    For a description of the 1988 Air Force and 1994 NASA reports, see “The Fall of TWA 800,” pp. 59, 60. 

  26. 26

    “Appendix D, Description of Events,” in Martin Shooman, “A Study of Occurrence Rates of Electromagnetic Interference (EMI) to Aircraft with a Focus on HIRF (External) High Intensity Radiated Fields,” April 1994, National Aeronautics and Space Administration, Langley Research Center, pp. 93-96 passim

  27. 27

    The fact that no warning alarm went off at the moment the autopilot disconnected has been widely cited in the press as evidence that the autopilot must have been disconnected by the pilots; but no such conclusion can be reached on the basis of an absent alarm. The Vienna Institute of Aerospace-Medicine and Space-Biology points out that the same electrical event that knocks out the autopilot can also disable the auditory alarm (“Faulty EgyptAir Device Suspected,” The Boston Globe, November 25, 1999). Similarly, Byron Acohido, a reporter for the Seattle Times, cautions that alarms are often disabled by a stray electrical signal (quoted in Jean-Paul Mari, “EgyptAir: Le Suicide Était Presque Parfait,” Le Nouvel Observateur, December 15, 1999, p. 43). 

  28. 28

    Conversation with Martin Shooman, October 31, l997 (two years before, and hence not referring to, the EgyptAir 990 accident). See Shooman, “Occurrence Rates,” p. 4, for an incident in which electromagnetic interference causes a problem with one engine; the autopilot compensates; the pilot then notices the engine problem and disconnects the autopilot; but now the compensatory actions of the autopilot have been interrupted and the plane enters a sudden dive. 

  29. 29

    Colonel Quisenberry, cited in “The Fall of TWA 800,” p. 60, notes 7-10. 

  30. 30

    Shooman, “Occurrence Rates,” p. 4 and p. 95 (case 40). Sudden dives, however, play a less prominent part in the NASA study than in the Air Force study because the NASA report is primarily about nonfatal electromagnetic events: it is based on accounts by living pilots, pilots who survived the incidents they are describing. As Martin Shooman points out, one reason electromagnetic interference is not adequately studied is that, in many cases, either its effects are short-lived and therefore may appear harmless, or they are fatal, in which case the pilot does not survive to inform us about it. One advantage of the Air Force study (and the three-year Pentagon study it motivated) is that it investigated accidents—such as uncommanded rolls and dives—that were fatal to the crews. 

  31. 31

    The Vienna Institute of Aerospace-Medicine and Space-Biology, in consultation with Austrian pilots with vast experience of the Boeing 767, has surmised that a possible cause of EgyptAir 990’s catastrophe was a “stabilizer runaway” that then caused an automatic disconnect of the autopilot. A stabilizer is a horizontal rudder on the tail that, as the name implies, usually keeps the plane level, but can, in the case of a runaway, tilt the nose down. (See “Faulty EgyptAir Device Suspected,” The Boston Globe, November 25, 1999.) The anomalous position of the elevators could have been part of the runaway, or could reflect the effort made by the pilots to override the stabilizer runaway by adjusting the elevators. 

  32. 32

    See “The Fall of TWA 800,” pp. 59-60. 

  33. 33

    Shooman, “Occurrence Rates,” p. 6. In this case both of the electronic engine controls were burned out, although in other cases engines stopped without suffering permanent physical damage. 

  34. 34

    Shooman, “Occurrence Rates,” Appendix, p. 93 (case 14). Effects on the engine other than a straightforward shutdown are also included in the NASA report: “H[igh] F[requency] transmitter interference with ana-log engine governor controls caused engine to surge. Varied with frequency…” (Appendix, p. 96, case 56). One oil pressure sensor inside an engine was affected by high-frequency transmissions (p. 96, case 54). 

  35. 35

    Recovered data boxes and cockpit voice recorders can greatly help to sort out which events in a given accident were initiated by the pilots and which events had some other source. But even these sources of information are not always decisive, and sometimes they are simply wrong. The data recorder can indicate events that investigators know did not take place: one altitude parameter on the TWA800 data recorder gave invalid readings and had to be ignored by investigators (NTSB “Flight Data Recorder (FDR) Group Chairman’s Factual Report,” February 15, 2000, revision, p. 5). Conversely, it can fail to announce events that investigators know did take place: in the case of Swissair 111, the smoke in the cockpit at 9:14 PM and fire soon thereafter are not registered on the data recorder, according to lead investigator Vic Gerden. If we did not have the exchange between the pilot and the air controllers, we would not know these key events had taken place. 

  36. 36

    Of the four events, only the last—the shutting down of the engines—currently has evidence suggesting that it was an act brought about by the pilots: the data recorders and voice recorder indicate that the instruction to shut down the engine, or at least partially shut it down, originated in the cockpit. Because this particular event took place thirty seconds into a dive of terrifying speed and angle (40 degrees, during one interval), it has been proposed by a number of analysts that the turning off of the engines may have been an accidental act—that the pilot may have been reaching for a different lever altogether (see Byron Acohido, “Pilot Could Have Confused Similar Switches on 767’s Control Panel as Plane Plunged,” Seattle Times, November 18, 1999) or the pilot may have tried to slow the engines down, an appropriate act in an emergency descent, but may have gone too far. (See C.J. Chivers with Matthew L. Wald, “EgyptAir Data Recorder Shows Engine Shutdown,” The New York Times, November 13, 1999, p. A7.) 

  37. 37

    The NTSB’s first public announcements about the recovered voice recorder (midday, Monday, November 15, 1999) described the conversations between the pilots as friendly, and reported that the crew had worked together cooperatively to try to undo the catastrophe. But by 5:00 PM on that same day, a different account began to emerge from the NTSB that dominated the press for many months and focused attention on three or four separate sentences spoken by the pilot or copilot.