His pressurised helmet connected to the Grasp, a Navy salvage tug, by a communications and oxygen-supply umbilical, Chief Petty Officer Oelhafen and his diving partner, Petty Officer Douglas Irish, descend gently to the ocean floor. Suddenly from the inky silence, piles of wreckage come into view, illuminated by their hand-held torches and the lights of a remote submersible already in the area. Jagged metal shapes rise from the sand. Swaying eerily in the deep- water currents are yards of bare wiring ripped from the jet.
The men know that their 90-minute dive will be hazardous. They must take care not to slice or tangle the umbilicals. Their orders, meanwhile, remain the same as on previous days: priority number one is to locate the bodies of passengers still missing. But then, almost immediately, the lights of the submersible settle on two bright orange boxes. The writing on their sides leaves no doubt as to the magnitude of the find: "Flight Data Recorder".
The first breakthrough of the investigation into the crash had finally arrived. "We've seen pictures of what they look like," Oelhafen later related. "We've been briefed on what to do." The two boxes, the cockpit voice recorder and the data recorder - they are orange in spite of their "black box" nickname - were raised and by 2am they were on board a Blackhawk helicopter on the first leg of their journey to Washington, DC, for analysis in the laboratories of the National Transportation Safety Board.
The tapes were found to be intact. What they revealed further compelled investigators to believe that TWA 800 had been destroyed by some kind of criminal explosion. The voices tape yielded 11-and-a-half minutes of routine cockpit conversation from the moment of take-off from JFK International and then they went dead. Just before the silence, however, there was a "brief fraction-of-a- second sound" that was strikingly similar to noises heard on the recorders aboard two planes downed by terrorist bombs in 1985 and 1988 respectively: an Air India flight off Ireland and PanAm flight 103 over Lockerbie. The very brevity of the noise will have been interpreted as further support of the bomb or missile theory but it is virtually impossible, even with the sophisticated computers available to the investigators, to determine whether the noise was a bomb. University researchers have attempted to analyse the sound spectrum of a bomb explosion to see whether it can be identified from cockpit voice recordings but so far without much luck.
Since the night of the crash, the FBI has been hunting locally on New York's Long Island and around the world for intelligence pointing to some terrorist conspiracy. Possible theories range from a bomb planted on board in a suicide or insurance-fraud stunt, to what now looks increasingly likely, a terrorist assault executed by a surface-to-air missile fired either from land or sea.
The less likely scenario - a massive mechanical failure on the aircraft - cannot be ruled out, however, until some actual physical evidence of an external attack is found. Such failures have never yet caused an instantaneous loss of electrical power and destruction of the plane. Typically, a structural break-up causes tremendous noise as the metal tears away from the structure. Such a sequence normally lasts several seconds at least, and would probably be recorded on the cockpit voice recorder. The flight data recorder, monitoring the movements of key control surfaces, and other data such as speed and heading, would be likely to show some unusual movements before the break- up.
The noise may be circumstantial evidence of a bomb or missile, but proof will come only from detailed examination of the wreckage, precious little of which has yet been recovered from the sea-bed. This was the case with the Lockerbie bomb. Confirmation that the brief noise recorded on the cockpit voice recorder was a bomb came only with evidence from the wreckage itself.
THE SEARCH, then, for such positive evidence continues this weekend at several locations. Investigators are still anxious for clues from the remainder of the plane's wreckage, more than 90 per cent of which remains under water. A new focus was provided by the discovery on Friday of two engines. They could prove critical in proving or disproving the missile theory.
Speculation about a missile began with the number of eye witnesses who reported seeing something streaking towards the airliner before it exploded. Additionally, still pictures from air traffic radars seemed to show an unexplained blip in the vicinity of the plane at the same instant. The blip was initially explained away as electronic interference but military officials remain haunted by the possibility that one or more shoulder- launch Stinger missiles which were smuggled to rebels in Afghanistan by the CIA in the 1980s may have fallen into the hands of a terrorist group hostile to the US.
Anxious to quash the missile talk (one version that refuses to go away suggests a US Air Guard plane in the area at the time might have accidentally downed the TWA jet), the Pentagon has emphasised that the aircraft was at an altitude of over 13,000ft when it met catastrophe, putting it at the very edge of the effective range of a Stinger. "The Army's Missile Command did some computer modelling, crunched all the data and determined it was not outside the realm of possibility, but it's also not likely," one officer told reporters. But reports yesterday suggested that a missile attack was now being favoured.
Lifting the engines to the surface for inspection is vital because Stinger missiles are heat-seeking; if one was indeed fired at the airliner it would almost certainly have struck at the rear of one of the engines as they roared to achieve cruising altitude. If an engine is found to bear evidence of a missile hit, then stage one of the investigation - what happened? - will be closed. Stage two will be the hunt for the perpetrators.
Pivotal also, meanwhile, is inspection both of the wreckage and of passengers' bodies by forensic scientists in search of residue that may have been left by an explosion. The leader of the forensic team is Tom Thurman, the chief of the FBI's explosives unit and a legend in the aviation accident community. It was Mr Thurman who made the breaks in determining that it was a bomb that downed PanAm 103 and who linked it to Libya. So far, no residue has been found and, as each day passes, the risk increases that any such evidence will be washed away by the ocean.
But if forensic tests fail to provide evidence of residue, there are still other trails of physical evidence that can be followed.
Divers have been making video tapes of the wreckage as it lies on the ocean floor so that the investigators can view the tapes, or watch the live relay from the remotely controlled submarines, to get their first look at the wreckage. Still more important will be analysis of the manner in which the parts of the plane broke apart. Investigators may be able to state with certainty that the damage has been caused, not by the impact of the plane on the water, but by an explosion.
Information emerging yesterday gleaned from debris position and radar scanner records points to an explosion that ripped out a portion of the right side of the aircraft and then ruptured the fuel tanks before triggering the fireball that sent it into the ocean nearly half a minute later.
Such evidence from debris, showing, for instance, whether parts of the skin were blown inwards or outwards, may also point to the nature of the blast. In the case of PanAm 103, the bomb hypothesis was reinforced by the discovery of damage to the tail caused not by the crash itself but by other parts of the plane blown out by the detonation. Even individual rivets can be examined to determine the direction in which they popped out and what kind of pressures were applied to them.
Using complex computer modelling calculating the terminal velocity of each piece of wreckage, the wind-speeds at the time, and the point in the air at which it began its free fall, it is possible to work out which pieces of the aircraft separated from the main structure first. This method is complicated by the fact that the wreckage struck the water, but there are computer models for sea currents too and, eventually, the investigators will be able to work out which bits came off the aircraft and when.
The sequence in which the plane broke up can tell investigators where to concentrate their examination of the wreckage. Attention is likely to be focused initially on the pieces that broke away first. As the jumbo's remains are assembled in a hangar the engineers' detective work on each recovered item continually narrows the focus. Explosions, fires, bombs and in-flight break-ups of planes all have characteristic signatures. An aviation-fuel vapour explosion will leave residues on the surfaces, and the forces that such explosions can exert have been quantified. High explosive detonates at a vastly greater rate than a fuel explosion, between 1,000 and 8,000 metres per second, much faster even than the high-speed debris of a disintegrating jet engine.
Under the microscope a bomb-damaged piece of aluminium or titanium, for example, will betray the velocity of the explosion and show hot-gas erosion, where the surface of the metal is seared by the extreme heat. Under an electron microscope there are a wide variety of features, such as pitting, curling, or fusing of the metal that are characteristic only of bombs. There will also be chemical residues of the explosive and its detonator. If a missile was involved, the investigators will find inward-pointing jagged edges on the remains of the part of the plane nearest the detonation or where the missile struck.
Still more arcane is the work that can be done inside the cockpit once it is located. If some kind of mechanical cataclysm did occur, all sorts of clues could present themselves. Investigators can, for example, examine the filaments in the bulbs of individual warning lights. If any were alight at the moment of the tragedy, the filaments will have been elongated slightly and will remain frozen in that state today.
THE bodies of the victims can reveal crucial evidence. One of the indecencies of such tragedies is that, for the purposes of a crash investigation, human remains are as much forensic evidence as the other items of wreckage. The victims may, for instance, display signs of eardrum damage or overpressure on the lungs which will point to particular types of explosion.
In 1967, for example, a British Comet airliner plunged into the Mediterranean, leaving only a few bodies and seat cushions on the surface, with the main wreckage too deep to recover. The then chief investigating officer of the UK's Accident Investigation Branch, Eric Newton, asked the pathologists to X-ray the bodies. Tiny fragments of metal were found in some of them which could only have entered the bodies at a velocity far higher than any impact force, or from the force of a fuel explosion. Similar fragments were found in the seat cushions.
This provided not just evidence, but positive proof that a bomb had exploded on the plane. Newton and his team positioned rods and wires along the trajectory on which the objects had entered the bodies and the cushions and were able to work out where the bomb must have been placed. Since then, the X-raying of bodies or body-fragments has become standard procedure in cases where bombs are suspected.
There are other characteristics of bomb explosions, among them the feature that first alerted investigators to the likelihood of a bomb having downed PanAm 103 over Lockerbie in 1988. A piece of the cargo container in which the bomb had been planted was embedded with parts of the radio-cassette player. The force required to embed the material in the cargo container wall was clearly far in excess of that which could be produced by anything other than a bomb explosion. Similarly, if a bomb is responsible for the TWA flight 800 catastrophe, it may well have embedded parts of its trigger mechanism in the aircraft wreckage.
AT THE National Transportation Safety Board's nondescript headquarters in Washington, DC, the examination of the various "black box" tapes goes on, with additional expertise being provided by British experts who took part in the PanAm inquiry. Repeated analysis of that final "sound" may eventually determine its source. The team will be able to compare how the sound was transmitted first through the air to the cockpit's microphones and second through the aircraft's skin. Because sound will have travelled faster via the skin, it should be possible to determine the distance of the blast from the cockpit.
And finally, back at the site of the crash nine miles off the southern shore of Long Island, Petty Officer Oelhafen and scores of his fellow Navy divers will continue probing the ghostly wreckage, and their search for the nearly 100 bodies that remain missing. To us it is still a mystery, but the suspicion that passengers and crew may have been fully aware of what was happening can only increase the horror of the task. "Nothing compares to this," remarked one diver, Andy Gliganick. "And I hope nothing in my life ever will compare to this."
The vital clues: what the investigators are looking for
Three scenarios are still being considered. Each new finding narrows the possibilities
n Probably the only cause of a such a rapid explosion would be a catastrophic engine failure: the investigators will determine whether the engines were running at the time of the explosion or when they fell into the sea
n In a structural break-up the noise of metal tearing would probably be recorded on the cockpit voice recorder. The flight data recorder would show unusual movements
n Evidence from the cockpit, including examination of the filaments of bulbs, could determine whether warning lights of failing systems had been activated
n The investigators will be comparing the sound heard on the cockpit voice recorder with previous known incidents caused by bombs
n A bomb would leave residue from accelerants on wreckage but some of this may have been washed off by the sea or sand on the sea bed
n Fragments of the bomb (or parts of the plane near to the explosion) would have penetrated deep into parts of the wreckage and into the bodies of the passengers
n Aircraft skin would have been blown outwards by detonation of a bomb from within the aircraft
n A missile would probably be heat-seeking: it would almost certainly have severely damaged a rear part of one of the engines. The missile could be triggered outside the aircraft by proximity to its target, in which case the wreckage would show damage from the outside
n The aircraft's altitude (13,000ft) puts it just within range of a ground-to-air missile
n Shrapnel penetrating bodies or wreckage would point to a missile attack: the direction of debris would be a decisive clue