TWA 800 and Electromagnetic Interference: Work Already Completed and Work that Still Needs to be Done

Editors’ note:The following is a Web-only supplement to a series of articles published by Elaine Scarry in The New York Review of Books.

In July of 1998, the National Transportation Safety Board announced that it would attempt to assess the strength of the electromagnetic environment surrounding TWA 800 the night it fell. In order to do this, it would enlist the help of two agencies: the Joint Spectrum Center, whose work is specifically dedicated to solving problems inside the military that arise from joint use of the electromagnetic spectrum, and the National Aeronautics and Space Administration (NASA), some of whose scientists study the ways electromagnetic interference can affect civilian planes and spacecraft.

A three-stage process followed, each stage building on the foundation laid by the stage that had preceded it. In August of 1999 the NTSB compiled and forwarded to the Joint Spectrum Center a list of sea- and aircraft in the vicinity of TWA 800. The Joint Spectrum Center, limiting itself to the transmitters on this list (as well as to its own list of fixed transmitters on land), then calculated the signal frequency and strength of the land, sea, and air transmissions at the surface of the plane. Its formal report, “TWA Flight 800 Electromagnetic Environment,” completed in January of 1999, identified forty-five transmitters whose signal strength at the accident site was higher than one volt per meter (the threshold figure the Safety Board asked them to use).1 NASA then took the Joint Spectrum Center figures and analyzed them in order to determine whether they were strong enough to have provided the ignition source for the central fuel tank explosion on TWA 800; NASA concluded they were not.2

Each of the three stages of work enumerated above broke new ground. First, as the Safety Board noted at its August 22-23, 2000, hearing, though it had previously looked at questions connected to electromagnetic interference in navigation and communication, this was the first time it had ever explored the way in which electromagnetic interference might act as an ignition source. Second, this was also the first time that the extraordinary resources of the Joint Spectrum Center had ever been called upon on behalf of commercial airline passengers3 : in the past, its resources had been dedicated exclusively to military planes4 or planes carrying high government officials.5 Third, as NASA’s formal report itself states, in the course of NASA’s work on behalf of TWA 800, their scientists developed a new method of analyzing electromagnetic interference from sources external to the plane that they hope will provide an important tool in future inquiries.6

These innovations have been appropriately saluted and placed in the public record by the agencies who carried them out. The article that follows below, while cognizant of these important achievements, will focus on the incompleteness of the work carried out on behalf of TWA 800. It will explain why the projects so far undertaken by the Joint Spectrum Center and NASA should—despite their virtuosities and complexities—be regarded as extensive preliminary studies rather than as final studies, why the work already accomplished needs to be supplemented by additional work. The fact that two other large passenger planes have fallen, or had their first sign of trouble, in the same region in which TWA 800 fell provides strong incentives for carrying out additional work, and those incentives have been examined in detail in Parts I and II of this article7 : what follows below specifies the additional research that the preliminary work on TWA 800 itself suggests should now be undertaken.

Both the Joint Spectrum Center and NASA have—in different ways (the first in oral statements, the second in written statements)—openly acknowledged the possible incompleteness of their work.

In their report entitled “Investigation of Electromagnetic Field Threat to Fuel Tank Wiring of a Transport Aircraft,” NASA scientists stated their conclusions that the HIRF, or High Intensity Radiated Fields, environment surrounding TWA 800 was not strong enough to have acted as an ignition source.8 According to their calculations, the signal strengths listed by the Joint Spectrum Center could have introduced at most 0.1 millijoule of energy into the fuel quantity wires running into the fuel tank, whereas 0.2 millijoules is the level required for ignition.9

NASA, however, appended a second supplementary study—entitled “Some Notes on Sparks and Ignition of Fuels”—which called into question the accuracy and applicability of the 0.2 millijoule standard when talking about RF (radio frequency) sources of electrical excitation. Its central argument is summarized in its opening abstract and restated in its conclusion:

There is little to suggest that data on ignition levels from high voltage sparking is particularly applicable to sparking from breaking contacts excited from an RF source.10

The second NASA report ends by specifying four avenues of future work that are needed in order to determine an accurate RF ignition standard. The report also observes that most of the work so far done on RF ignition appears to have been carried out in Britain, Germany, and other European countries, and worries, “This fact may indicate that there is not a cadre of American investigators with much background in RF ignition problems.”11

Though the NASA report talks about one very specific “unknown” in the region of electromagnetic interference (the correct standard for ignition from an RF source), this open acknowledgment of what is so far “unknown” turns up again and again throughout many other areas of research on electromagnetic interference, whether the particular report is being issued by the FAA,12 by the Air Force,13 by manufacturers of shielding, or by some other sources.14 To imagine, therefore, that the door to additional research can be crisply closed on this subject seems incompatible with the generally acknowledged inscrutability of electromagnetic interference, as well as the very specifically acknowledged inscrutability of the accurate standard for RF ignition.

In addition to the question about NASA’s research that NASA itself raises, other questions may be relevant to pose. The scientists carrying out the first of the two studies, “Investigation of Electromagnetic Field Threat to Fuel Tank Wiring of a Transport Aircraft,” repeatedly describe their approach as taking “the worst case” so that not all variations of a given scenario have to be tried: given the scores and scores of possibilities that would have to be contemplated in order to assess every possible variation, this “worst case” shortcut seems extremely sensible. But is it the case that the “worst case” instances really represent the worst case? (This is not a rhetorical question: it may well be that the study does in fact stay exclusively with the most negative possibility. But the only way to make the worst-case status of the worst case visible is to set forth multiple cases and to learn which one of them is “the worst.”)

Three ways in which the study may fall short of a “worst case” design will be specified here. According to the Joint Spectrum Center, at the moment that TWA 800 lost its transponder and all other electrical power (8:31:12 PM), forty-five transmitters in the region were capable of producing signals at the accident site with a field strength that was one volt per meter or higher. The moment selected for the study is the moment when all the clocks (and all other electrically driven instruments on TWA 800) stopped; it is an appropriately selected moment to stop all external action as well and see where each transmitter in the region was. But because the Joint Spectrum Center report also indicates the bearing of each transmitter, it can in many cases be determined whether, in the sixty seconds prior to the clock-stopping moment, a given transmitter was closer or further away and therefore whether its field strength was higher or lower.^15 Despite the fact that the Joint Spectrum Center figures made it possible to calculate field strengths both at 8:31:12 and in the minute prior to the visible onset of the accident, NASA chose to work only with the 8:31:12 figures.

At the clock time of 8:31:12, the highest field strength figure was the transmitter at Brookhaven Laboratory: its emissions on the surface of the passenger plane were 32.6 volts per meter. NASA therefore took this 32.6 volts per meter as the “worst case.” According to NASA’s figures, once this Brookhaven signal passed into the passenger cabin, it was capable of introducing into the wires running into the central fuel tank 0.097 millijoules of energy, a figure which NASA usually rounds off to 0.1 millijoule. This 0.1 millijoule figure—half of the 0.2 millijoules that currently accepted standards require for ignition—was in NASA’s judgment the worst-case electromagnetic insult to TWA 800 from an external source.

But was it? The Navy P3 plane produced a field strength of 23.8 volts per meter on the surface of TWA 800 at the 8:31:12 clock time when the two planes were separated by a distance of 2.93 miles. But fifteen seconds earlier the P3 had come within one mile of TWA 800. (The one-mile figure is uncontroversial: it reappears in statements by the FBI, by the Pentagon, as well as in the visual and verbal record of the NTSB.16 ) That means that fifteen seconds earlier, the P3 was producing signals that had a field strength of almost seventy volts per meter on the accident plane.17

As the field strength of the Navy P3 changes, so too does the level of energy it is capable of coupling into the central fuel tank wire. While the field strength varies inversely with the distance (2.93 miles), the power and energy levels vary inversely with the square of the distance (2.93 x 2.93). According to NASA figures, the Navy P3 when at a distance of 2.93 miles was capable of introducing 0.032 millijoules into the wire running into the central fuel tank. That means at a distance of one mile it would introduce 2.93 x 2.93 x .032 millijoules or close to 0.3 millijoules (0.27 millijoules).

The reader will note that this figure—0.27 millijoules—places it above the 0.2 millijoules level required for ignition. (The one-mile distance, it is crucial to state again, is not speculative: it is not arrived at by saying “What if the P3 had passed within a mile of TWA 800?” The P3, according to the NTSB radar maps, as well as verbal statements throughout the investigation made by the NTSB, FBI, and Pentagon, did pass within a mile). Even using the conservative 0.2 millijoules standard (the figure whose accuracy has been challenged by the second NASA study), a figure significantly beyond that 0.2 requirement has been found.

Does this mean that the ignition source for the TWA 800 accident has been found? It could mean that. At the very least it means that a potential ignition source has been found and that the door to future research on electromagnetic interference has swung open.

To determine whether the P3 could have acted as the ignition source, important additional calculations would have to be carried out.18 It might be useful to answer many nonnumerical questions as well. It is important to recall, for example, that the P3 itself reported many electrical problems, radio problems, and control surface problems when it returned to its base in Brunswick, Maine, that night. Could some or all of the malfunctions been induced by a High Intensity Radiated Field?19

H[igh] F[requency] 1 failed to couple entire flight

#3 sh[i]p gauge failed intermittently throughout the entire flight

#3 autofeather will not arm

Copilot master ICS box, no U[ltra] H[igh] F[requency] 1 or 2 lights

H[igh] F[requency] amp causes constant transmitter off light

ECA [Electronic Countermeasures Action] # 12 card…causes bearing #2 needle to be off 30 degrees

I[dentification] F[riend] F[oe] transponder bad, no mode C

M[agnetic] A[nomaly] D[etector] R-32 black pen is inop[erative;] fuse checks good

Mode C inop[erative] during entire flight20

#2 powerlever 36 degree switch inop[erative], will not autoshift with powerlevers forward of 36 degrees21

Rudder trim has to be turned several turns before it engaged and trimmed the aircraft. Happened several times throughout flight22

This list constitutes a high number of electrical (or potentially electrical) malfunctions. It is only a partial list of the many problems the P3 pilots reported when they returned to their air base the night TWA 800 fell. In the forty-two days prior to July 17, the crews of this particular P3 reported a total of eight malfunctions: that means that twice as many malfunctions happened on this single night of July 17 as happened in the entire six weeks leading up to that night.

At what point in its flight did these electrical malfunctions begin? 23 If the P3 first suffered these problems close to the time TWA 800 fell, then both planes may have been the victim of some third transmitter (it is conceivable that a high-powered military transmitter seeking to put the relatively immune P3 in its field of vision inadvertently illuminated both the P3 and TWA 800 at the moment when the two planes crossed paths).24 If instead the Navy P3 suffered its electrical problems only long after TWA 800 fell, it remains a candidate for the ignition source of the TWA 800 explosion. What is clear is that the crew of the P3 themselves should be asked specifically for whatever pertinent information they can supply concerning the crash of TWA 800. So far nothing in the record indicates that this has been done.25

But let us proceed to a second area in which a question can be raised about whether the NASA study successfully achieves a “worst case” design.

The Joint Spectrum Center had presented NASA with a list of forty-five signals whose field strength on the surface of TWA 800 ranged from one volt per meter to 32.6 volts per meter (and, as noticed a moment ago, up to seventy volts per meter fifteen seconds earlier).

NASA then took these figures indicating the strength of the signal on the outside surface of the plane and (applying their new method of numerical computation) calculated what the strength would have been once the signals came in through the windows in the passenger cabin; they then calculated what the strength of the signals would have been once it passed from the passenger cabin onto the wires (running inside the walls and flooring beside the passenger seats) that travel into the central fuel tank.26

But this seems a somewhat circuitous path to the particular wire under investigation (the fuel quantity system indicator wire running into the central fuel tank). Is it the most direct way that High Intensity Radiated Fields could have coupled onto the wire? Since this passage involves attenuation of the signals, is this the path that would indicate the severest impact of the signals?

The answer to these questions may be “yes.” It may well be that this indirect path to the fuel quantity indicator wire does represent the worst case, and that the various physical properties of the passenger cabin mean that it would—despite attenuation—result in the largest figure. But how do we know in advance what the “worst case” is unless several cases are looked at and the one with the worst result chosen?

An alternative way that the HIRFs on the exterior wall of the plane could conceivably have come into direct contact with the fuel quantity indicator system wire is through the wheel well. (The wheel well is the cavity into which the landing gear retracts after takeoff and prior to landing.) The National Transportation Safety Board’s exhaustive study of the interior of TWA 800 had in fact expressed a concern about the uneven levels of shielding on the plane’s wiring. The wire bundle that travels into the central fuel tank at one interval runs through the wheel well where it is unshielded for a distance of ten feet. Because of its location on the underside of the plane, the wheel well is almost immune to lightning strikes, and for this reason the wire was permitted to be unshielded. But though the wheel well is immune to lightning, it is no more immune to radio and radar signals than any other aperture, or port of entry, into the plane.27

In November of 1998, the FAA issued an Airworthiness Directive applicable to all 747 models requiring replacement of “all of the FQIS wiring outside of the fuel tanks and surge tank with shielded wiring.”28 This action will provide protection for the ten feet of formerly unshielded wire running through the wheel well (as well as for hundreds of other feet of FQIS wire running through other sections of the plane29 ). The directive states the purpose of the new requirement:

To prevent electrical transients induced by electromagnetic interference (EMI) or electrical short circuit conditions from causing arcing of the fuel quantity indication system (FQIS) electrical wiring or probes in the fuel tank(s), which could result in ignition of the fuel tank(s).

TWA 800 prompted the new requirement for the shielding of this wheel well wire in Boeing 747s; but TWA 800 was not, of course, itself the beneficiary of the new requirement, and the question remains whether it could have been here—in the wheel well—that signals from various transmitters could have coupled onto the wires.

While the NTSB report indicates that the FQIS wire is vulnerable to the outside world during its ten-foot run through the wheel well, NASA overlooked the wheel well. Its report assumes that external HIRFs could only arrive at the fuel quantity indicator wire by the indirect, two-step process of coming in the windows, before then passing through the interior walls to the wires. It in fact states that HIRFs can only indirectly, not directly, reach this particular wire.30 This is especially puzzling because though NASA does not see the wheel well as vulnerable to the outside world, it does see the wheel well as the place where any electromagnetic signals inside the cabin (whether from HIRFs that have arrived there through the windows or from passenger-carried computers and cell phones) can most adversely affect the wire.31

How much energy couples onto the wire is in part determined by the relation between the wavelength of the impinging radiation and the length of the wire, or what is called effective antenna length, and can, as NASA notes, only be determined experimentally rather than computationally since it varies so much with frequency.32 What strength would the Brookhaven signal at the frequency of 2875 MHz (that was 32.6 volts per meter on the outer surface of the plane) have once it directly coupled onto the wheel well wire; what strength would the Riverhead transmitter signal in the 1294.6 MHz range (17.9 volts per meter at the outer surface of the plane) have once it directly coupled onto the wheel well wire? What strength would the forty-three other signals in different frequencies have been? Perhaps the answer is that the levels would be lower than the levels NASA has already found in the two-step path; but it seems worth finding out for sure.

NASA experimentally tested different frequencies (that it assumed to be coming from inside the cabin) and found one very surprising result, a result listed as the third of the three major conclusions to this first study. Though HIRFs could not have acted as the ignition source (conclusion one), and though passenger computers and cell phones could not have acted as the ignition source (conclusion two), low-frequency signals33 illuminating defective wire could cause discharge (or sparking) events by introducing as little power as 0.17 W (conclusion three). But they determined that this could not explain the fuel tank explosion because passenger cell phones and computers don’t operate in this frequency, and though HIRFs do operate in this frequency (among many other frequencies),

analysis revealed that the whole TWA-800 aircraft could not have coupled more than .015 W of power in this frequency range from the surrounding EM environment at the time of the accident.34

But if the .015 W power level has been arrived at only by the two-step process of indirect illumination by HIRFs (first having it pass through the cabin window, then having it illuminate the wires running through the walls), then it would seem relevant to determine the power level at this frequency by direct illumination in the wheel well.35

A third question about NASA’s “worst case” design involves the multiplicity of signals in the TWA 800 environment. In a complex electromagnetic environment, can the “worst case” be identified by isolating any one solitary emission—whether Brookhaven’s 32.6 volts per meter, or the Navy P3’s 23.8 volts per meter field strength, or the P3’s 69.7 volts per meter field strength at fifteen seconds before the catastrophe manifested itself, or Riverhead Air Force antenna’s 17.9 volts per meter, or any one of the forty-two other emissions so far not enumerated here? One of the most common observations made about electromagnetic interference is that it is hard to reproduce in the laboratory even when scientists are fairly confident that it has taken place in the field; and one of the most common explanations given for this difficulty in replication is that the laboratory often tests one frequency at a time, rather than reproducing the actual situation in the field where multiple frequencies are present and may be interacting.

Multiple signals within a single frequency (the two readings at the 2745 MHz frequency from the two radars at Islip provide a simple instance) can add or subtract from one another depending on the degree of phase alignment of the two signals. The formula for stating the worst-case outcome for signals at the same frequency (in the highly unusual case where the two signals are perfectly aligned) requires that their respective strengths be added together (in the 2745 MHz frequency, the signal strength would be 7.6 + 7.4 or 15 volts per meter); a more usual formula states the average outcome of interaction by squaring each number, adding the squares, then finding its square root (in the 2745 MHz frequency, the revised reading would be 10.6 volts per meter). While signals at the same frequency can amplify, or diminish, one another’s effects, so can signals in adjacent frequency bands, and so, too, can frequencies that occur at harmonic multiples of one another.36 If TWA 800 suffered from electromagnetic interference, it is possible that a single frequency from a single transmitters was the cause. Or it is also possible that a complicated interaction between some or all of the forty-five signals registering higher than one volt per meter was the cause.37

The FAA’s 1997 Flight Standard Information Bulletin about High Intensity Radio Fields38 had warned that a “fast changing R[adio] F[requency] environment” outside a plane can cause power to jump from high-voltage wires to low-voltage wires at some place inside the plane where the two are bundled together. Just such a jump from high- to low-voltage lines inside TWA 800 had been contemplated by the NTSB as a possible avenue along which inappropriate levels of power had been carried into the fuel tank,39 even though the NTSB had not connected that concern with an external radio frequency environment or with the FAA advisory on High Intensity Radiated Fields. Does the Joint Spectrum Center’s list of signals from forty-five fixed and mobile emitters constitute “a fast changing R[adio] F[requency] environment”?40

Does the list of forty-five emitters, once supplemented with many other potential emitters omitted from the original study, constitute such a “rapidly changing radio environment”? It is to the subject of the omitted transmitters that I now turn. Of the three stages of work—

  • the NTSB compilation of transmitters;

  • the Joint Spectrum Center’s computation of the strength of the transmitter signals once they left the site of transmission and traveled to the surface of the plane; and

  • the NASA analysis of whether these transmission figures could have served as an ignition source

—it is so far exclusively the third stage of work that has been contemplated here. NASA explicitly states its own recognition of the need for additional work in a second report exclusively devoted to the problem of determining an accurate RF ignition standard. I have attempted to suggest a number of other ways in which NASA might be asked to undertake additional research: the P3’s field strength fifteen seconds prior to the visible onset of the catastrophe appears to create an energy level significantly above the 0.2 millijoule level currently understood to be required for ignition, and calls attention to the fact that the emission levels of transmitters in the two-minute period prior to the accident need to be looked at; the wheel well could be tested as a possible site of direct illumination by HIRFs and compared with the two-stage path NASA has already examined; and the interaction of multiple signals could be tested. Decrease in craft distance, more direct paths of illumination, and the interaction of multiple signals have been introduced here one by one, but all of them need to be studied together.41

But as we now back up from the third stage of work (NASA) to the second stage (the Joint Spectrum Center), and from the second stage to the first (the NTSB), we will see that the incompleteness of the first stage of work puts all later stages in jeopardy.

Just as NASA openly acknowledges, by its second report, the incompleteness of the work it has so far carried out on RF ignition, so too does the Joint Spectrum Center openly state a crucial caveat about its own report: it acknowledges that its own accuracy is dependent on the accuracy of the list of transmitters provided by the NTSB.42 If the NTSB list omits sources, the Joint Spectrum Center’s report omits sources, and in turn NASA’s conclusions understate the hazard.

It will be useful to contemplate briefly the Joint Spectrum Center’s own impressive resources, and then see how those resources can be thwarted by inaccurate information forwarded to it by the prior stage of investigation.

When the National Transportation Safety Board asked the Joint Spectrum Center to give a technical portrait of the electromagnetic envelope surrounding TWA 800 at the moment its catastrophe began, the NTSB was asking the Joint Spectrum Center to carry out a task requiring great virtuosity, but one the Center was well practiced in carrying out. Since the 1960s when it came into being, its central work has been to solve problems that arise from shared use of the electromagnetic spectrum, or what it refers to in its mission statements as “E3” problems, a shorthand for Electromagnetic Environmental Effects. Its facilities in Annapolis are split into two buildings on opposite banks of the Severn River. The headquarters—located beside a nature reserve that also contains the remains of a once-spectacular antenna array used for submarine communication throughout the twentieth century—is jointly staffed by members of the Navy, Air Force, and Army who rotate responsibility for the Joint Spectrum Center’s leadership. Located here too, behind locked doors, is the highly classified library of radio and radar frequencies, the most complete record of transmitters anywhere in the world.

If the Joint Spectrum Center had set out to tell the Safety Board the strength of all antennas and sensors at the site of transmission, the classified library alone would have provided the answers. But they sought to answer a more difficult question (and the only question that mattered). They wanted to determine the strength of the signals not at the site of transmission but at the site of reception—the site of reception being TWA 800.

For this task, the data bank of frequencies and signal strengths located at the Joint Spectrum Center’s headquarters was crucial but insufficient. Across the Severn River near the Naval Academy is the Joint Spectrum Center’s second, much larger building: it contains laboratories and offices for four hundred scientists trained in electrical engineering, computer science, and physics. Technically the civilian organization housed here is a part not of the Joint Spectrum Center (or any other military or governmental body); it is part of the large, civilian, nonprofit Illinois Institute of Technnology Research Institute that has branches in several states. However, this particular branch of IITRI, called IITRI Annapolis, carries out research projects for only a single entity, the Joint Spectrum Center. Its contract with the Joint Spectrum Center began when the Joint Spectrum Center began, in the middle 1960s, and continued uninterrupted throughout the Seventies, Eighties, and Nineties (during most of which time the Joint Spectrum Center was called the Electromagnetic Compatibility Analysis Center). It is, in fact, the longest continually running single contract in the United States government.

In August 1998, the Joint Spectrum Center and its partner, IITRI Annapolis, took on the task of determining the exact field strength of each radio or radar signal in the region of the TWA 800 accident not at the moment it left its land, sea, or airborne antenna but at the moment it arrived at the passenger plane making its way through the summer sky.

The National Transportation Safety Board clearly made an effort to be accurate in the list it forwarded to the Joint Spectrum Center. The list contained revisions in what had earlier been specified as the distances of various craft from the accident plane. The Aegis cruiser USS Normandy had always earlier been described by NTSB, FBI, and Navy officials as 180-200 miles away; its distance was now corrected to 156.3 miles. The Normandy had also been repeatedly described as the ship (other than the Coast Guard cutter Adak) closest to the accident, a description that turned out not to have been accurate. The NTSB’s list now specified two surface ships that had not been included in any previously announced FBI or NTSB listing. The SS Halyburton, an Oliver Perry Class Navy frigate, was located 130 miles east-northeast of TWA 800, and the Navy supply ship USN Seattle was located 43 miles to the west.43

But the list was still starkly incomplete:

Certainly Omitted Craft One: Moving on the water three miles south of TWA 800 was a ship traveling at thirty knots. It is visible on the NTSB radar data. Because it is so close to both TWA 800 and the Navy P3 plane, it has been an object of steady interest throughout the inquiry. Both the FBI and the NTSB have at every moment throughout the investigation certified that this object was indeed there (given the radar data, any other position would be almost impossible); but they have, with almost equal consistency, acknowledged frank ignorance about what it is.44 If efforts have been made to ascertain its identity, those efforts have not been described to the public.

The investigation into the accident by former FBI assistant director James Kallstrom entailed, according to his own statements, direct inspection of 371 vessels passing though New York Harbor45 and scrutinizing the records of 20,000 vessels that passed under one of three Suffolk County bridges in a three-month period.46 Yet another ship—one large enough to show up on radar data, sailing not in New York Harbor or under one of the Suffolk County bridges but out in open ocean waters close to the plane—has been left uninterrogated and unidentified.

The Joint Spectrum Center has provided meticulously calculated figures for signals arriving at TWA 800 from Navy and Coast Guard ships 5 miles, 40 miles, 130 miles, and 156 miles away. The usefulness of those meticulous calculations is seriously compromised if the signals arriving from a ship 3 miles away have been omitted.

The spectrum of possible safety or harm is wide: the ship may have been a fishing boat that carried no high-powered transmitters; or it may have been a pleasure ship carrying no transmitters; or the ship may have been a second Aegis cruiser. (We know from the Joint Spectrum Center calculation that the Normandy—even at a distance of 156.3 miles—produced a field strength of 3.77 volts per meter at the outer surface of TWA 800; and from this we may estimate that an Aegis cruiser 3 miles away would produce a field strength of 196 volts per meter on the accident plane. So, too, we know that at a distance of 156.3 miles, the Normandy was capable of introducing into the passenger cabin .017 millijoules of energy47 ; thus we also know that at three miles it would be capable of introducing 45.97 millijoules (this would be considerably above the 0.2 millijoule standard required for ignition).48 The classified field strength levels of the antennas on our telemetry ships (we have two on each coast; they are used to track cruise missile tests) may be even higher.49 But why, rather than stating the possibilities, do we not have a simple direct identification of the ship as we have for the 371 ships the FBI personally boarded and the 20,000 ships whose papers it inspected?

Certainly Omitted Craft Two, Craft Three, and Craft Four. Three other ships—located respectively two, three, and six miles northwest of TWA 800—also appear on the NTSB radar data but have never been identified, and therefore could not be included in the study of transmitters carried out by the Joint Spectrum Center.50 The NTSB forwarded to the Joint Spectrum Center the radar display of planes and ships located within a few miles of the accident plane; but while the nearby planes carried identifying labels on the radar picture, the ships bear only a label indicating their speed (twelve-fifteen knots in two cases, twenty knots in the third). If the Joint Spectrum Center knew the identity of the ships, it would then—drawing on its library—also know the transmitters carried by those ships, and could then go on to calculate the strength of the signals produced by those transmitters at the accident site.51 But the first step in this sequence is the single step that the otherwise highly competent Joint Spectrum Center cannot carry out: it has no legal authority to inquire into the identity of the ships; that task must be carried out by the legally authorized agency for accident investigation, the NTSB.

The list of ships the NTSB forwarded to the Joint Spectrum Center includes ships as far away as 217 miles, 243 miles, 252 miles, 260, 269, 278, 304, and 347 miles. (None of these ships were incorporated into the Joint Spectrum Center’s study since they are all beyond the radio line of sight.) Such listings, along with the listing of ships in coastal ports, give an aura of completeness, even overcompleteness. But how can a list be complete if no ship within a ten-mile radius of the accident is listed, with the exception of the Coast Guard ship Adak,52 whose motions were widely reported by the press at the time of the accident, whose identify has therefore been consistently known to both the public and to accident investigators, and whose transmitter signals were therefore included in the Joint Spectrum Center study?

Possibly Omitted Craft Five: Navy documents (obtained by Freedom of Information and forwarded by this author to the NTSB) 53 indicate that there were two P3s in the Long Island region on the evening of the accident. One is designated as being in the New York area: this is the already long-identified P3 that came close to intersecting TWA 800’s path moments before the fall and is included in the Joint Spectrum Center and NASA studies. The second P3 is designated as being in the W-105 region and is not included in the study.

Does the absence of the second P3 from the Joint Spectrum Center study mean that the NTSB never bothered to inquire into its location? Or does it instead mean that upon inquiry, the NTSB learned that, contrary to the Navy document,54 no second P3 was in the W-105 region? The NTSB needs not only to track down answers but to make certain the answers are publicly known. Just as the NTSB’s attention is obligatory, not discretionary, so the public’s attention in the case of the death of fellow citizens is obligatory, not discretionary. Information needs to be given to the public at large so it will understand when it has successfully discharged its own obligations to fellow citizens who have been mysteriously killed or injured.

None of the many craft listed by the NTSB appears to be situated inside the exercise zones W-106 and W-105, portions of which are within ten miles of the crash. Does that mean W-105 and W-106 were empty on the evening TWA 800 fell? Or does it instead mean that the NTSB has included only craft in the civilian air lanes and sea lanes? The place on the map where W-105/ W-106 begins is marked with clear lines, giving the appearance that these regions can be crisply segregated off from the narrow corridors of civilian waters. But electromagnetic signals do not stop traveling through the air at the place on a map where a line is drawn. If there were craft inside W-105 and W-106 they must be included in the study.

Possibly Omitted Crafts Six, Seven, Eight, Nine, Ten, Eleven, Twelve, Thirteen, Fourteen, Fifteen, Sixteen, Seventeen, Eighteen, Nineteen, Twenty, Twenty-One, Twenty-Two, Twenty-three, Twenty-four, Twenty-five, and Twenty-Six: Beginning in August 1999, the NTSB made available to the public its radar data from a geographical area larger than the twelve-by-sixteen mile region shown throughout its week-long December 1997 accident hearings. An independent researcher from the University of Florida and another from the University of Maryland each separately transcribed the data in the band just beyond the region that had earlier been shown and found images of more than twenty ships, some inside W-105 and others moving together from a position outside the warning area into the warning area in the fifteen minutes following the fall of the plane.55 The transcribed data also shows two planes south of TWA 800, one flying toward W-105 and the other circling into and out of the warning region.

The NTSB has not itself stated in public whether these findings are accurate. If they are not accurate, the NTSB needs to show its own method of transcription and the pictures that result. If they are accurate, then the NTSB needs to identify the craft and forward them to the Joint Spectrum Center so that it can supplement the work it has already carried out and forward the new figures to NASA. Portions of W-105 and W-106 are only ten miles from the accident: it should go without saying that the use of powerful transmitters in this area could significantly change the picture of the electromagnetic environment surrounding TWA 800.

Incompletely Described Crafts Twenty-Seven and Twenty-Eight: The twenty-seventh and twenty-eighth craft involve imprecision (rather than an omission) but imprecision so large it comes close to being a stark omission. Two Air Force planes—a C-141 and a KC-10—are acknowledged to have been in the area. But neither their distance from, nor bearing in relation to, TWA 800, is specified. The distance is listed as twenty-five miles, but is followed by a note indicating that this is an “assumed distance separation.” How much is being “assumed” in the twenty-five-mile figure? Give or take three miles? Nine miles? Twenty miles? Power levels of radio and radar transmissions fall off at a rate of the square of the distance as one moves away from the object; phrased the other way round, power levels increase at a rate of the square of the distance as one approaches the object. Whether the C-141 is twenty-five miles away or seventeen miles away or seven miles away makes a large difference in the Joint Spectrum Center’s field strength calculations. Equally crucial is the bearing of the two planes which the NTSB simply dismisses as “unknown.” (Bearing lets us deduce how close to the accident plane the other two planes were one minute prior to the crash.)

It is hard to puzzle out why, after securing the help of the Joint Spectrum Center—the one laboratory in the whole country with a technical expertise so comprehensive and so long practiced it could perform the exact calculations needed to see if the plane was vulnerable—the NTSB then decided to throw all precision to the wind by “guesstimating” that the C-141 and the C-10 were—“shall we call it twenty-five miles?”—away from the place where 230 people died.

The following craft, then, need to be identified and added to the Joint Spectrum Center’s otherwise virtuoso study, and added in turn to the research carried out by NASA that builds on the Joint Spectrum Center computations:

  • one ship located three miles south of TWA 800 whose existence is certain;

  • a second, third, and fourth ship located two to ten miles northwest of TWA 800 whose existence is certain;

  • a P3 in the W-105 military area whose existence is indicated in a Navy document but whose existence has not yet been verified by the NTSB;

  • a group of more than twenty ships and two planes in, or near, the W-105/W-106 area whose existence is indicated by radar pictures transcribed by two independent researchers from data provided by the NTSB, but whose existence has not yet been verified by the NTSB itself;

  • one US Air Force C-141 and one US Air Force C-10 whose distance, bearing, and altitude need to be precisely specified.

If the NTSB has reasons why none of the listed craft need to included in the Joint Spectrum Center and NASA studies, it should state those reasons publicly.56

This list is limited to those craft that are, in varying degrees, already “in evidence,” either because they are in the radar maps or a Navy document establishes its existence. Some of our military equipment is, of course, radar-evading; its whereabouts will only be ascertained when the men and women operating in the area are called upon to describe events in the region.57 As I have argued from my initial article on TWA 800 in April of 1998, these men and women are—even more than the high-powered radars and transmitters—the eyes and ears of the country.

The reports and documents pertaining to the NTSB’s inquiry into TWA 800 reveals an investigation that is—in the scrupulous care of its analysis of the plane itself, a masterpiece of precision: the documents map soot patterns on every square inch of the plane’s exterior; every wire, every instrument, every air passageway, every circuit breaker, has been scanned centimeter by centimeter and described in clear language. Members of the research teams even figured out the spins and spirals that small pieces of metal and furniture inside the plane went through as the plane fell. But compared to the tour de force internal mapping, the external mapping of the environment is a blur of partially accurate, partially approximate, half-complete, and wholly absent information.

Incomplete in the case of TWA 800, the reconstruction of the external electromagnetic environment has only barely begun in the cases of Swissair 111 and Egypt Air 990. Will we only become persuaded of the need to look at the external environment if a fourth, a fifth, and a sixth plane fall? Shall we spare ourselves the exhausting labor of inquiry and wait for the environment itself to eliminate all doubt? But surely we are required to act while we still stand in the midst of uncertainty, long before events themselves coerce our conviction.

Acquiring information about the external environment is an obligation, not an option, an obligation that should have been perceived as absolute even when there was only a single catastrophe in this geographical region.

Air Controller: TWA 800, Center.

[No Response]

Air Controller: TWA 800, Center.

[No Response]

Air Controller: TWA eight hundred. If you hear Center, ident[ify yourself]

[No Response]

That obligation grew heavier when, twenty-six months after this first catastrophe, the second followed, its earliest signs of peril occurring in the very geography where its sister plane had fallen.

Air Controller: Swissair 111, Boston.

[No response]

Air Controller: Swissair 111. Cleared direct to Bradd.

[No response]

Air Controller: Swissair 111. Swissair 111. Hear Boston Center. Contact Boston one two eight point seven five, one two eight point seven five. If you hear Boston, ident[ify yourself]

[No response]

Fourteen months after the second catastrophe came the third, the plane having crossed into and crashed within the military zone the two earlier planes had been skirting when they flew the Bette route.

Air Controller: EgyptAir 990. New York Center.

[No response]

Air Controller: EgyptAir 990. New York Center.

[No response]

Air Controller: EgyptAir 990. If you copy New York Center, squawk one seven one two and ident[ify yourself]

[No response]

How much time do we have before a fourth plane—taking off from JFK airport and traveling east over the ocean—fails to answer its radio call?

It is a question that deserves further inquiry.

  1. 1

    The study’s major authors were Richard DeSalvo, Martin Macrae, and Douglas Hughes. In the NTSB’s accident inquiry documents, the report is Docket No. SA-516, Exhibit No. 9A. Addendum 2, “Concerning Electromagnetic Environment.”

  2. 2

    At its August 22-23, 2000, final hearing on TWA 800 in Washington, D.C., the National Transportation Safety Board stated that (as a result of the Joint Spectrum and NASA studies) they considered electromagnetic interference an “unlikely” source of TWA 800’s ignition.

    However, at three different points in the hearing, three different investigators (Robert Swaim, Scott Warren, and Chairman Jim Hall) took care to note that though High Intensity Radiated Fields were judged “unlikely” to have caused ignition in this case, they were capable of causing ignition in other cases under different circumstances. (Bernard Loeb, Aviation Safety Director, also repeatedly stated that causes ruled unlikely in the case of TWA 800 could act as ignition sources in other cases. But he was not specifically directing attention to electromagnetic interference.)

  3. 3

    The Joint Spectrum Center’s resources (as will be elaborated below) include a comprehensive library of the world’s transmitters and over four hundred scientists whose work is exclusively dedicated to analyzing and eliminating conflicts in the use of the electromagnetic spectrum. At present there does not appear to be any equivalent resource in the civilian sector.

  4. 4

    During United States intervention in former Yugoslavia, for example, military pilots repeatedly found that voice communications were interrupted by voices speaking in Spanish. The Joint Spectrum Center eventually tracked the problem to the unregistered use of a microwave antenna in a South American country. Though the people using the antenna only intended to transmit information locally from the bottom of a mountain to its top, the signals (which happened to be aligned with an orbiting NATO satellite) had a repeated effect thousands of miles away where they jeopardized the communications and hence the safety of military pilots. The mystery was solved not by analyzing the content of the fragmented transmissions but by mapping an ellipse of the area in which plane transmissions had been affected and then determining the locations from which signals arriving in that ellipse could have originated. (Conversation with Commander John Mahoney, Joint Spectrum Center, Annapolis, Maryland, November 1998.)

  5. 5

    In addition to the many investigations it has carried out on behalf of military persons, the Joint Spectrum Center analyzed the possibility that electromagnetic interference could have affected the plane carrying Secretary of Commerce Ron Brown that crashed near Dubrovnik, Croatia, on April 3, 1996. The plane was a CT-43A (the Air Force version of a Boeing 737); all thirty-five people on board died. (Letter to author from Jim Hall, Chairman of the NTSB, July 8, 1998).

  6. 6

    Called MoM (Modal Analysis/ Method of Moments), the tool provides a “new…numerical simulation technique” for modeling the changes in energy levels as that energy moves into and out of large geometries, such as the passenger cabin of a Boeing 747. (J. Ely, T. Nguyen, K. Dudley, S. Scearce, F. Beck, M. Desphande, C. Cockrell, “Investigation of Electromagnetic Field Threat to Fuel Tank Wiring of a Transport Aircraft,” NASA Technical Paper 2000-209867, Hampton, Va.: Langley Research Center, March 2000, pp. vii, 13, 14, 18-20, 25-26, 32- 35, 39). NASA scientists report that they have verified their new method by testing it against other numerical methods that leave more factors unknown and that require large amounts of computer time.

  7. 7

    Part I of this article, published in The New York Review of September 21, 2000, looked at the fall of Swissair 111 and identified eight features it shares with TWA 800 (repeated features that may provide important clues that the external environment is playing a part): 1) the two planes took off from the same airport; 2) they took off on a Wednesday at 8:19 PM; 3) they traveled along the Bette route; 4) they both had their first signs of trouble in the same region of air space between twelve and fourteen minutes into the flight; 5) they both suffered an electrical catastrophe; 6) they both suffered a catastrophe whose cause remains mysterious, even after years of rigorous inquiry; 7) they both flew during a week when extensive military exercises were being conducted; 8) they both flew when certain specific transmitters (submarines, the Navy plane P-3s) appear to have been in the region.

    Part II, published in The New York Review of October 5, 2000, described five reasons why electromagnetic interference needs to be looked at in the case of EgyptAir 990. First, as the former head of the Joint Spectrum Center has pointed out, such reconstruction needs to become a routine part of accident investigations. Second, if a plane falls in a military zone (where powerful transmitters are sometimes in use), there is even more reason for including such a review. Third, if a plane falls in or near a region where other unsolved accidents have either ended (TWA 800) or begun (Swissair 111), the obligation grows still greater. Fourth, if major features of the plane’s fall (in the case of EgyptAir 990, the disconnecting of the autopilot, the sudden dive, the anomalous action of the elevators, the shutting down of the engines) are individually compatible with electromagnetic interference as described by the 1988 Air Force study, the 1994 NASA studies, or other major studies in other countries, the obligation to scrutinize the environment is still stronger. Fifth, if the record of civilian air controller tapes indicates that the controllers themselves were at any point either surprised or worried about the plane’s routing (and the controllers along EgyptAir 990’s route expressed such unease at many points), the pressure toward scrutiny mounts.

  8. 8

    The vocabulary used for describing electromagnetic signals varies depending on whether a point in space, a two-dimensional area, or duration through time is at issue. Electric field (expressed in volts over some distance such as a meter) is a measure of force on charges placed at a point in space. Power density (expressed in watts over an area such as a square meter or square centimeter) is a measure of the power flowing across an area. Energy or work (expressed in joules or a subunit such as millijoules) indicates the power delivered over a period of time.

    The Joint Spectrum Center calculated the electric field strength on the outside of the plane (using the measure volts per meter); the NASA study calculated the energy of those signals once they traveled through the windows into the passenger cabin (using the measure of millijoules); the NASA study, at various points, also calculated the power density of the signals once they coupled onto the wires running beside the passenger cabin (using the measure of watts).

  9. 9

    Investigation of Electromagnetic Field Threat,” Section 3.7, “Conclusions for External Analysis,” p. 39.

    Several scientists whose expertise is in the area of electromagnetic interference point out that it would be helpful if NASA could indicate the range of error associated with its 0.1 millijoule calculation.

  10. 10

    Franklin A. Fisher, “Some Notes on Sparks and Ignition of Fuels,” NASA /TM-2000-210077, pp. 1, 34.

  11. 11

    Fisher, “Some Notes on Sparks and Ignition of Fuels,” p. 34. (This observation may leave out the extensive work on this question carried out in the United States automotive industry.)

  12. 12

    This past year, for example, the FAA set forth a new set of shielding requirements on the 767-400 that place much higher levels of shielding on this new plane than on almost any other civilian plane (including the closely related 767-300 involved in the EgyptAir 990 accident). The shielding requirements are prefaced as follows:

    It is not possible to precisely define the HIRF [High Intensity Radiated Fields] to which the airplane will be exposed in service. There is also uncertainty concerning the effectiveness of airframe shielding for HIRF. Furthermore, coupling of electromagnetic energy to cockpit-installed equipment through the cockpit window apertures is undefined” (Federal Register, Vol. 64, No. 139, July 21, 1999, Proposed Rules, p. 39097).

  13. 13

    For example, Carl E. Baum, senior scientist at Air Force Research Laboratory, Kirtland Air Force Base, and internationally recognized theorist on electromagnetic interference, was quoted in my original article on TWA 800 (The New York Review, April 9, 1998) describing the elusiveness of the phenomenon:

    It is hard enough to calculate the fields on the exterior of a complex system such as a ship, missile, aircraft, tank, etc…. As one goes to the interior and encounters thousands of wires going to numerous black boxes, antennas, etc. through various cavities and other strange structures the calculation problem gets quickly intractable. Even if one thought he could calculate such a complex problem, he would likely miss important signals because the system in the field is often different from the drawings by various changes for the convenience of the operators” (“From the Electromagnetic Pulse to High-Power Electromagnetics,” in Proceedings of the IEEE, Vol. 80, No. 6 (June 1992), p. 802).

    Baum’s statement also, of course, calls attention to the difficulty of the calculations that both the Joint Spectrum Center (that calculated the fields on the outside of TWA 800) and NASA (that calculated the fields on the inside of TWA 800) undertook to carry out.

  14. 14

    In addition to the question the second NASA report raises about the standard for RF ignition, NASA’s first report several times acknowledges the incompleteness of our knowledge about various aspects of electromagnetic interference. For example: “The radiated field characteristics of small cavities such as the Center Wing Tank bays and the test chamber are not well understood” (“Investigation of Electromagnetic Field Threat to Fuel Tank Wiring of a Transport Aircraft,” p. 82).

  15. 16

    For citations from FBI, NTSB, and Pentagon spokespersons, see my first article on TWA800 in The New York Review, April 9, 1998, p. 65 n.40; p. 67 n.42, and map p. 73. Several official statements place the P3 directly above TWA 800, as does a Navy report obtained through Freedom of Information that documents the Navy’s investigation to make certain no ordnance had fallen off the plane. That inquiry would presumably have been undertaken only if the P3’s position relative to TWA 800 was such that an object falling from it could intersect the path of TWA 800 below.

  16. 17

    The figure 69.73 volts per meter is arrived at by multiplying the field strength (23.8 volts per meter) by the ratio of the distance (2.93 miles over 1 mile or 2.93).

  17. 18

    The 0.27 millijoule figure tells us the strength once the signals had traveled from the outside of the plane to the passenger cabin inside; NASA scientists point out that signals lose additional strength as they make a second passage from the cabin interior to the wire running into the fuel tank. As will be elaborated below, however, there are paths other than the one NASA proposes by which the signals can get to the central fuel tank that do not require a two-step process.

  18. 19

    The pilot complaints that follow were earlier printed in the July 16, 1998 issue of The New York Review. (“An Exchange on TWA 800,” which printed my June 17, 1998, letter to Chairman James Hall, p. 56. This material on the P3 was first obtained by a citizen who made a Freedom of Information request to the Navy; in response to my April 9, 1998, New York Review article on TWA 800, he contacted me and forwarded to me the Navy materials he had obtained.)

  19. 20

    This is the P3’s transponder and seems to suggest that the reason the P3 was flying without the transponder was that it didn’t work, rather than that the pilots intentionally turned it off to obscure their altitude. The repair mechanic examining the transponder after the plane returned to the base, however, reports that the transponder was working: “Could not duplicate [complaint] on the deck. Syst[tem] checked 4.0 on ground test.”

  20. 21

    This is one of several pilot-reported problems that the repair mechanic states could not be duplicated on the ground: “Could not duplicate gripe, RPM switch checks good on low power turns.” One possible explanation for the discrepancy between pilot’s and mechanic’s reports is that the pilots suffered electromagnetic interference to those systems during the flight that did not leave permanent damage that could later be detected on the ground.

  21. 22

    The pilots speculate that the cause is mechanical (“Rudder trim wheel appears to have slipped inflight when trimming to the right”). But here again the mechanics on the ground cannot duplicate the problem, suggesting that the problem may not be mechanical in origin: “Performed Op[erational] check of rudder trim tab system…. Op[erational] check 4.0, could not duplicate gripe.”

  22. 23

    Of the problems listed, only the inoperative transponder is officially on record as having begun prior to the moment that the P3 crossed TWA 800’s path. (On the Boston FAA air transcript, one air controller states to a second controller that the P3 is not using its transponder, and three times—at 8:20, 8:22, and 8:23 PM—expresses concern that it will soon be crossing through the departure route out of JFK airport). In two other items listed above, the pilot uses the word “entire” flight; yet because the P3 entire flight was 6.7 hours long, the word “entire” could be used for the 6.2 hours that followed the moment when (30 minutes after its own takeoff) it crossed TWA 800’s path.

  23. 24

    If both the P3 and TWA 800 suffered the onset of their respective electrical difficulties at the same time, it is still conceivable that the P3 is the source: that is, it could have used a piece of equipment that affected its own systems at the very moment that it adversely affected TWA 800.

  24. 25

    A Navy inquiry (according to one document received under Freedom of Information) checked the plane itself to make certain no ordnance had fallen off (which, had it fallen, might have hit the commercial plane flying below). But nothing in the Navy or NTSB records documents the crew’s understanding of their own electrical difficulties, nor their statements about transmitters they were using or which they knew other craft to be using at the time, or any other insights they might be able to provide about the region of the accident.

  25. 26

    Here is where NASA applied its new “Modal Analysis/Method of Moments code”: they made a small one-foot model of the passenger cabin, found the electric field strength, then carried out computerized extrapolation (in some frequencies requiring over a hundred hours of computer time and in some frequencies thousands of hours) to determine the power and energy levels.

    NASA was, as part of the same study, scrutinizing the signals from PEDs, or passenger-carried electronic devices, such as cell phones and computers. By calculating the strength of the HIRFs once they passed through the windows, the HIRFs and PEDs could be compared from the same starting point.

  26. 27

    NTSB TWA 800 Investigation, Attachments to Systems Group Chairman’s Factual Report, Exhibit 9C, ATA 28-41-00, “TWA Wiring Investigation Summary: Fuel Tank Wire Routing, Center Tank FQIS Component Test,” p. 197.

    When the NTSB was enlisting NASA’s help, did they forward this document to them (as well as all other documents in their investigation materials directly relevant to the question about electromagnetic interference NASA was asked to answer)? The NTSB investigation of TWA 800 involves thousands of pages so it would seem to place a difficult burden on NASA to ask them to find all the key passages themselves.

  27. 28

    FAA, Airworthiness Directive, Large Aircraft, 98-20-40 Boeing, Amendment 39-10808, Docket 97-NM-272-AD. The Advisory Directive gives Boeing thirty-six months to carry out the wiring replacement on all 747-100, -200, 300, -SP, and -SR series airplanes.

  28. 29

    Unlike the wire in the wheel well, however, most other sections of the wire would be shielded against signals arriving from outside the plane because of the aluminum fuselage.

  29. 30

    Investigation of Electromagnetic Field Threat to Fuel Tank Wiring of a Transport Aircraft,” p. 41.

  30. 31

    NASA predicted computationally and then experimentally confirmed that the greatest problem occurred within ten meters of the wheel well. Immediately after passing through the wheel well the wire enters the center fuel tank; so it is here that the set of low-impedance wires get braided together and a set of high-impedance wires get arranged in “a daisy chain” fashion immediately before traveling into the central fuel tank (“Investigation of Electromagnetic Field Threat to Fuel Tank Wiring of a Transport Aircraft,” pp. 52, 53, 85).

  31. 32

    The basic unit for frequency is hertz (named after the German physicist Heinrich Hertz) which is one cycle per second, or megahertz (abbreviated MHz) which is one million cycles per second.

    While the coupling energy is affected by frequency, it also depends on whether the electric field is parallel or perpendicular to the fuselage (called “polarization”) and the angle at which it hits the fuselage (called “angle of incidence”).

  32. 33

    The 2-10 MHz signals are at the lower end of the HF (high frequency) band.

  33. 34

    Investigation of Electromagnetic Field Threat to Fuel Tank Wiring of a Transport Aircraft,” pp. viii, 142.

  34. 35

    See note 41 below for a description of the way the power level of these 2-10 Mz signals might be affected not only by a more direct path of illumination, but by variations in the distance of the transmitter and by the interaction of multiple signals.

  35. 36

    It is also the case that under certain conditions, new frequencies (called “beat” frequencies) can be introduced that are either the sum or the difference of two already existing frequencies.

    As Martin Shooman (the author of the 1994 NASA study on HIRFs) points out, especially relevant to the TWA 800 accident might be “beat” frequencies in the 2-10 MHz range, since the 2000 NASA study states that arcing can occur at these lower frequencies if the plane has damaged wiring (Conversation, October 13, 2000).

    A 1995 NASAstudy of electromagnetic interference on space missions came to my attention after I published my articles in The New York Review. It opens by citing a famous case of electromagnetic interference from beat frequencies that occurred during the development of the Saturn launch vehicle when one of its key receivers (responsible for engine cutoff, arm, and destruct commands) received spurious signals (R.D. Leach, author, and M.B. Alexander, editor, “Electronic Systems Failures and Anomalies Attributed to Electromagnetic Interference,” NASA RP1374, Marshall Space Flight Center, July 1995, 2.1.1). This 1995 report describes many of the crashes I have mentioned in my New York Review articles (five Blackhawk helicopters, one F-111, one German fighter plane) but also includes crashes and near crashes about which I was previously unaware. Among these are: (1) the crash of an F-16 near a Voice of America transmitter that apparently helped initiate the FAA’s attention to the problem of High Intensity Radiated Fields; (2) the crash landing of a drone or RPV (remotely piloted vehicle); (3) an engine “over speed condition” on an Apache helicopter that “could have resulted in a double engine failure”; (4) the near crash in 1993 of a commercial DC-10 as a result of electromagnetic interference from a passenger-carried device; and (5) many instances of “phantom commands” on satellites or planes or cars. The report states that NASA’s space vehicle program has had few in-flight problems from electromagnetic interference because they spend so much time in the development stage working to eliminate those problems; the report ends by urging that such vigilance be continued.

  36. 37

    It is conceivable, for example, that many of the forty-five transmitters are in use in the Long Island region as planes fly through there, but that a specific set of three almost never used at the same time were in use the nights that TWA 800, Swissair 111, and EgyptAir 990 fell. This is one reason why, as I have stressed since the April 1998 New York Review article on TWA 800, it is crucial to gather complete information about all transmitters in use.

  37. 38

    For a description, see my article “The Fall of TWA 800” in The New York Review, April 9, 1998, p. 73.

  38. 39

    The fuel quantity indicator system is an instance of a low-voltage wire bundled together with high-voltage wires: according to NTSB documents, it “is routed in bundles with almost 400 other wires carrying electrical power of 5 to 192 volts” (Systems Group Chairman’s Factual Report, Docket No. SA-516, Exhibit No. 9A, p. 10). In the NTSB’s final hearing on TWA 800, Robert Swaim described the possibility of this jump from high-voltage to low-voltage wires incited by a short circuit (August 22, 2000, Washington, D.C.).

  39. 40

    When the FAA wrote this advisory, what level of abruptly changing RF environment did they have in mind? Were they picturing 245 emitters rather than the 45 emitters at the TWA 800 accident site, or the same number of emitters but ones whose field strength would be much higher than those at the accident site? We can be confident that they were not expecting a commercial passenger plane to fly into a full-fledged electronic warfare exercise since planes that do this have extremely high levels of shielding: even a Navy plane not designed for electronic warfare but designed to land on a carrier is shielded to 2000 volts per meter.

    A clearer answer to this question than formerly existed is beginning to emerge. The NASA study reports that while it was carrying out its work, another research group was carrying out a study of the HIRF environment at airports (where there are extremely high concentrations of transmitters because so many planes are gathered in one place, and where the airport radars are nearby). It then shows figures indicating the airport HIRF figures are much higher than the figures surrounding TWA 800 at the moment of its fall (“Investigation of Electromagnetic Field Threat to Fuel Tank Wiring of a Transport Aircraft,” pp. 16, 17).

    But clearer labeling needs to be provided before the skeletal information about airports provided by NASA’s terse summary can be understood and assessed. The airport figures for “normal peak” strengths, for example, are so high (thousands of volts per meter) that they appear to be in drastic violation of national standards for permissible level of emissions. The American National Standards Institute/IEEE recommendation for “limits of maximum exposure to radiated fields” specifies ten milliwatts per square centimeter of surface for a duration of six minutes (or 194 volts per meter for six minutes). This safety standard has been formally adopted by OSHA. It is highly unlikely that airports would permit levels that could adversely affect the people working all day on or near runways; so it may be that the figures indicate the field strength at the surface of the transmitter rather than in the extended environment. A second way in which the chart is difficult to read is that it shows all frequencies simultaneously in operation (though it almost surely intends to suggest that when a specified frequency is in use, it operates at the specified average and peak strengths). The TWA 800 figures with which is contrasted (and which are placed on the same chart) represent a particular moment, not all possible frequencies that are ever transmitting in the airspace seven miles south of East Moriches in a twenty-four-hour day as hundreds of civilian and military planes transit through the area.

    A 1996 British Civil Aviation Authority report analyzes transmissions at Heathrow, Gatwick, Warlingham, Greenford, and Clee Hill airports. Though the study worries about figures obtained in certain frequencies, the levels do not appear to be close to the airport study NASA alludes to: for example, a level of 10 volts per meter (in frequency range between 80 MHz and 1000 MHz) is described “as a severe electromagnetic radiation environment.” The highest readings on their charts of the five British airports are 50 and 100 volts per meter (again, the report NASA alludes to shows thousands of volts per meter in some frequencies). The British report was undertaken in order to bring British airports into conformity with European Directive 89/336/EEC for Electromagnetic Compatibility, which became mandatory on January 1, 1996 (S. Nensi, Electromagnetic Compatibility Assessment of Large Air Traffic Services Equipment, CAA Paper 96007, London, 1996, pp. 23, 24, 25).

    A 1975 United States study of Palmdale (CA), Oakland (CA), and Islip (NY) airports found that electromagnetic emissions exceeded the required standard but in most cases did not adversely affect equipment in the area if it was properly shielded. The study measured emission levels of various transmitters, and also tested the vulnerability of airport equipment to surrounding transmissions. As in the British study, the levels being talked about do not resemble the power levels in the 1999 airport study alluded to in the NASA report: some pieces of equipment are described as susceptible at as low as one volt per meter; others malfunctioned when radiated with spikes of twenty-five or fifty volts per meter (Margaretta V. Stone, Radio Frequency Interference (RFI)/ Electromagnetic Interference (EMI) Measurements Air Route Traffic Control Centers, AD-A008 439 , FAA, April 1975, pp. 25, 49).

  40. 41

    An example is provided by NASA’s discovery that signals in the low end of the HF (High Frequency) bandwidth—between two and ten MHz—could, in the presence of defective wiring, introduce a spark with as little as 170 milliwatts of power. Let us see how 1) change of distance, 2) direct path of illumination, and 3) multiple-frequency interactions all work together.

    Passenger-carried devices do not operate in this two-ten MHz range; and none of the external HIRF signals calculated by the Joint Spectrum Center as having a field strength greater than one volt per meter included this frequency. But NASA, to its credit, went back to the Joint Spectrum Center to inquire the strength at the accident plane of any transmitter signals in the low end of the HF band (no matter how small their strength). It turned out that six transmissions from the Navy P3 were in this frequency band. NASA, following its supposition that external signals could only arrive at the fuel quantity indicator wires by the indirect route of first traveling into the passenger cabin, calculated how much power each of the six would introduce into the passenger cabin and found that the highest had a power of 14.8 milliwatts—far less than the 170 milliwatts required for ignition in the presence of defective wire (“Investigation of electromagnetic Field Threat to Fuel Tank Wiring of a Transport Aircraft,” p. 138).

    But when the Navy P3 was a mile away (rather than 2.9 miles away) the power level of this same signal (2.74 MHz) would be 125 milliwatts, which begins to look much closer to the 170 milliwatts needed for ignition when defective wiring is present. And if we now test interactions between this one signal in the 2.74 MHz frequency and the five other signals in nearby frequencies (all between 6 and 7.5 MHz), and if we further test a direct path of illumination (the wheel well is one such path, but there are others), the resulting “worst case” may be worse than the 125 milliwatts already arrived at, and far worse than the 14.8 milliwatt figure NASA calculated.

    The Navy P3 transmissions described in this note are in the two-ten MHz range and are distinct from the P3 transmissions in the 8500-9600 MHz range described earlier.

  41. 42

    This point was several times explicitly made to the author by two different members of the Joint Spectrum Center (November 1999, January 2000); and it is also implicit in the written report which contains radar maps of equipment from the accident area, some of it identified and some of it unidentified.

  42. 43

    Deck logs for these ships provided to the author by the Naval Historical Center (Naval Yard, Washington, D.C.) show that throughout the day and evening of TWA 800’s fall, the Halyburton was moored at Newport, Rhode Island, and the Seattle was moored at Naval Weapons Station Earle, New Jersey. (Neither of these two ships, therefore, can account for the large military ships several witnesses saw moving along the Long Island shore on the afternoon of the accident.) It is odd to include moored ships in a report that, as outlined below, excludes ships on the open sea within several miles of the accident plane.

  43. 44

    National Transportation Safety Board, Public Hearing, Monday, December9, 1997, Morning Session. See also in NTSB inquiry documents, “Airplane Performance Study,” Docket No. SA-516, Exhibit No. 13A, p. 5.

  44. 45

    The 371 were vessels that had entered or exited New York harbor within twenty-four hours of the crash (James Kallstrom, news conference concluding FBI probe, as summarized by Fred Kaplan, “FBI Finds No Evidence of Missile or Bomb in Blast,” in The Boston Globe, November 19, 1997, p. A14).

  45. 46

    News conference concluding FBI probe, as summarized by Benjamin Weiser, “In Graphic Simulation, FBI Tries to Show Jet’s Fiery End,” in The New York Times, November 19, 1997, p. A34. Commenting on the thoroughness of his inquiry, Kallstrom said that the letters FBI stand for “Federal Bureau of Total Investigation” not the “Federal Bureau of the Obvious.”

  46. 47

    NASA, “Investigation of Electromagnetic Field Threat to Fuel Tank Wiring of a Transport Aircraft,” p. 16, Table 3.3.1-1, “Dominant Emitter Characteristics.” See column 4, row 12.

  47. 48

    The figure 45.97 millijoules is arrived at by taking the ratio of the distance (156 miles over 3 miles, or 52), squaring it (52 x 52, or 2704), and multiplying the power at the original location by that figure (2704 x 0.017 millijoules = 45.97 millijoules).

  48. 49

    I would ordinarily abstain from designating what the most powerful ships could have been were it not for the odd occurrence in the final NTSB hearing of anecdotal evidence (see following note). If eyewitness accounts are to be used, they should be eyewitness accounts from people who saw boats off the southern coast of Long Island on the day of the accident, not those who saw boats on some randomly selected later day.

    One witness, a former Navy man, saw a ship on the coast on the afternoon of the accident that he identified as either an Aegis or a Spruance class destroyer (Robert Davey citing Dean Seward in The Village Voice, March 3, 1998). What he saw cannot be accounted for by either the SS Halyburton or the USN Seattle since, according to deck logs, both those ships were in port all afternoon. Another citizen on the southern coast of Long Island saw a ship that had a radar dome. The only US ships with radar domes are telemetry ships. These are ships with extremely powerful transmitters, though their strength is classified.

    Ordinarily deck logs of United States ships are fairly easy to obtain from the Navy Ship Library. In the case of one of our East Coast telemetry ships, the USS Redstone, obtaining records has not been possible. The librarian in charge of deck logs for telemetry ships has spent seven months searching for the ship’s July 17, 1996, record and can’t find it. Not only librarian Nancy Barr but (according to Barr) “the whole East Coast” has been, in response to the author’s requests, searching for this ship’s deck logs. “Someone” who cannot be named, according to Ms. Barr,remembers seeing the ship berthed in Florida that day because of engine trouble. But it would be more helpful to have the actual deck logs. Deck logs for the country’s other telemetry ships have been obtained by the author, and the ships were not near the northeast waters.

  49. 50

    During the December 1997 public hearing in Baltimore, the NTSB had expressed troubled ignorance and sober puzzlement about the ships: Chairman Jim Hall and radar expert John Clark agreed that on the basis of its thirty-knot speed, it was fair to surmise that the radar target south of TWA 800 was a ship (rather than, for example, a helicopter).

    By the August 2000 final hearing, however, the unknown ships were treated casually and anecdotally by NTSB member Charles Pereira (perhaps his anecdotes were spontaneously offered, and were not a formal part of his report; they in any event seemed a departure from the standard of evidence and care usually taken at the NTSB).

    He held up a radar image of the region from a randomly selected day three years later to show that on this day there were also craft in the sky and sea (something which it probably never occurred to anyone to doubt). He made no effort to identify the radar marks on either day. The mere presence of radar marks from the accident day and a randomly chosen day three years later was somehow supposed to indicate that unidentified craft near TWA 800 were incapable of affecting the plane adversely.

    He reported to those assembled at the hearing that he sometimes went boating in Long Island Sound “in a similar type of area” and had sometimes personally observed fishing boats. Using this as evidence for why there is no need to identify the ships actually present on the day of the accident would be the equivalent of supposing that there was no need to examine the fuel in the plane that day because one of the investigators happened to use fuel processed by the same company in his automobile once when he went on a picnic.

    The introduction of this anecdotal evidence from days other than the accident was especially startling given that actual eyewitness reports by citizens who saw large military ships on the afternoon of July 17, 1996, along the southern coast of Long Island were not mentioned.

    No one on the four unidentified ships reported seeing the breakup of the plane or any of the falling debris, despite the fact that they were much closer to both events than all but one of the 736 witnesses who (according to investigator David Mayer and Chairman Jim Hall at the NTSB final hearing) saw some part of TWA 800’s fall.

  50. 51

    This statement includes the possibility that they are ships carrying no transmitters and that their power level at the accident site would therefore be zero.

  51. 52

    The Adak is listed in the Joint Spectrum Center report as five miles north-north east of TWA 800, a bearing and distance presumably forwarded to the Joint Spectrum Center by the NTSB. (However, the NTSB’s radar data seems to show no craft five miles north-northeast of TWA 800.) One seemingly possible solution to the double puzzle (the absence of the Adak on radar; the presence of an unidentified thirty-knot target) would be to conclude that the thirty-knot target is the Adak. (The Adak does, in fact, sometimes travel at a speed of thirty knots; and the close pattern of radar hits allows for the possibility that we are seeing the path of the ship as it first moved south, then retraced its own path back toward the accident to begin to rescue.) But it must be the case that the thirty-knot target simply is not the Adak, since the NTSB and FBI would have every reason to present the public with a solution to this mystery, and they have steadily declined to do so.

  52. 53

    The New York Review, July 16, 1998, p. 55.

  53. 54

    It seems unlikely that the announced P3 flight was canceled since the Navy document itself lists flights that were canceled that night, and the P3’s flight is not among them.

  54. 55

    Kelly Patricia O’Meara, “New Radar Data, New Questions,” Insight on the News, The Washington Times, September 20, 1999, pp. 24 and 25. The transcriptions made by the University of Maryland physicist are reproduced in Insight. The transcription made by Thomas Stalcup, a doctoral candidate in physics at the University of Florida, were presented in a televised news conference (C-Span, September 17, 1999).

  55. 56

    As observed earlier (note 2), Safety Board members stated during the final hearing that though they found electromagnetic interference an unlikely ignition source in the case of TWA 800, it could act as an ignition source “in other circumstances.” If additional craft carrying high-powered transmitters were in the vicinity of TWA 800, that would constitute a set of “other circumstances” that should be studied.

    Additional work on electromagnetic interference is especially appropriate since the short circuit hypothesis—the potential cause of the ignition the Safety Board found most likely (or, as Bernard Loeb put it, the cause that they were not able to set aside as unlikely)—actually did not appear during the final hearing to have any more direct physical evidence supporting it than did electromagnetic interference. (Investigators openly stated that they had no direct evidence of a short cirtuic or even a probable location.) There were, however, according to Robert Swaim, two crucial pieces of circumstantial evidence supporting the short circuit hypothesis: one minute and fifty seconds before the catastrophe, the pilot said to his copilot, “Look at that crazy fuel flow indicator…”; the data recorder also registered an anomalous condition called “dropouts” in which background electrical humming several times disappears.

    But it is crucial to notice that these same two events are equally supportive of the electromagnetic interference thesis. In fact, in my original article on TWA 800 (April 9, 1998), I had cited the pilot’s statement to his copilot about the fuel flow indicator as possible evidence. In a letter replying to my article, Chairman Jim Hall stated: “The examination of the flight 800 flight data recorder and the cockpit voice recorder did not indicate any unusual signals before the end of the data” (Letter, April 21, 1998, reprinted in The New York Review, July 16, 1998). The Safety Board has now changed its position on the potential significance of the pilot’s words, even if the board members have forgotten their earlier dismissal, and overlooked the fact that those words (as well as the dropouts) provide equal support to both the possibility of short circuit and the possibility of electromagnetic interference.

  56. 57

    We have many planes and ships that are designed to evade radar. For example, by 1993 thirty-nine of our ships were equipped with “Operation Bandit,” and seventy Aegis cruisers were scheduled to be outfitted by 1996. Operation Bandit entails the use of R[adar]A[bsorbing]M[aterial] that covers the decks, superstructures, and masts. Portholes are covered with “gold-impregnated plastic film, to keep radar signals from entering compartments, where they might be reflected from internal corners.” The result is that these huge ships have a smaller radar signature than a decoy (Norman Friedman, The Naval Institute Guide to World Naval Weapons Systems, 1007-98, Naval Institute Press, 1997, p. 549).

    Expendable decoys and jammers are another entire category of equipment whose possible presence at the accident site can only be learned about by drawing on the direct knowledge of the men and women who were in the region. For a description of the 1x5-inch expendable jammer, called POET, see The New York Review, April 9, 1998, p. 71.

    Often unreflective judgments are made about what power level can and what power level cannot introduce harm to a plane. But even a radar-related material that is completely passive—chaff—has been shown to have unexpected electrical consequences. (Chaff, according to the Chambers Science and Technology Dictionary, consists of “radar-reflective strips or particles dispensed from aircraft, missiles, or guns to confuse radar.”) I will cite unexpected qualities of chaff here as an example of the importance of looking at what was actually at the accident site, rather than assuming there is no need to inquire because we can already be confident that everything conceivably there is harmless.

    In 1985, 65,000 residents of San Diego suddenly lost electrical power when chaff dropped by Navy planes over the sea was carried by unexpected winds into the San Diego region causing “rolling power outages.” Five days later, 3,000 people lost power in the same region when several chaff canisters accidentally fell off a military plane. The chaff was of different kinds and lengths: some the size of glitter was found on the ground; some in long delicate strands hung, like angel’s hair, across power lines; much of it was shiny but small dull filaments of carbon clung, as though magnetized, to the windows of houses and cars. Some samples consisted of aluminum-coated fiberglass (a common form of chaff); other samples consisted of metal-coated wool; still other samples were of an indeterminate composition (San Diego Union Tribune, Jan. 15, 17, 28; July 27, Sept. 28, Dec. 8, 1985; Los Angeles Times, Jan. 12, 15, 16, 18, Sept. 27, 1985; Conversations with San Diego residents and employees of San Diego Power and Light, March 1999).

    Prior to the San Diego accidents, the military had used chaff only to produce false radar pictures (the power blackouts were a surprise). After the San Diego accidents, the military started intentionally using chaff to knock out electricity and all the systems that depend on it. Chaff made of carbon fibers caused power outages throughout Baghdad by being introduced into the city’s electric power- generating plants and onto power lines at the start of US military actions there (The Washington Times, October 21, 1991; Aviation Week and Space Technology, December 7, 1992). More recently, carbon-fiber chaff was used by NATO to cause widespread electrical power outages throughout Belgrade (The Times (London), May 4, 1999; The Herald (Glasgow), May 10, 1999, p. 9).

    Chaff should be looked into in the case of TWA 800 for the general reason that all forms of radar-related equipment should be looked into, and also for the following specific reasons:

    1) The passengers on TWA 800 were, according to news reports, covered with luminous metal flakes; rescuers lifting bodies from the dark water during the first night described those bodies as having a strange glow due to a coating of metal particles; observers lifting, or holding them, during daylight hours also stated they were coated with shiny flakes. Various explanations were provided in the press. Some news stories cited Coroner Charles Wetli, who attributed the metal flakes to the plane’s insulation; other news stories reported that theatrical glitter had been carried in the cargo hold, but gave no NTSB or FBI attribution. But no description of, or explanation for, the flakes appear in the NTSB documents.

    2) “Metal fragments” were found embedded in the sealant on the outside seams of one panel of TWA 800’s fuselage (“Structure’s Group Chairman’s Factual Report of Investigation,” Docket No. SA-516, Exhibit 7A, p. 28).

    3) Fiberglass and carbon filaments were found in the plane’s interior. “Transparent glass fiber” covered with opaque material lay inside ducts where air arrives from the engines after it has been treated by air conditioners (Report 97-1C0089, in Appendix III, “Fire and Explosion Groups Factual Report,” Docket No. SA-516, Exhibit No. 20 C, p. 18). “Fibrous material” was found in the circuit breakers for the scavenge pump relay and the circuit breaker for the reserve fuel transfer valve (“Contracted Laboratory Documentation,” SA-516, Exhibit 9-B, pp. 55, 57).

    4) A scattered field of radar clutter appears in the vicinity of TWA 800 during its fall and for the next twenty minutes. At the NTSB Public Hearing, NTSB radar expert and Deputy Director for Research and Engineering John Clark said that though the images might be thermal signatures from the explosion, no solid account could be given. The section on radar in the NTSB inquiry documents includes a page from an FAA handbook,Fundamentals of Primary and Secondary Surveillance Radar, that explains how false radar images lasting twenty or thirty minutes can be produced from chaff (SA-516, No. 13A, “Airplane Performance Study”).

    5) The Navy inquired into the possibility that ordnance might have accidentally fallen off the P3 and found that it had not. But the Navy document registering this inquiry does not record any attention to the possibility that a chaff canister might have mistakenly fallen from the plane (as it had in the second San Diego incident in 1985). One knowledgeable military observer has stated that P3s rarely carry or use chaff.