But even without aggression, injuries from "accidents" are an occupational hazard of routine stunts. Bruce Stephens, former director of animal behaviour for Sea World in San Diego and now a consultant to marine parks, says orcas have hurt him dozens of times but he still doesn't hesitate to jump into the water. "Any person who has trained these animals has been thumped, bumped, bruised, bitten and otherwise abused over the course of time," he told Nancy Cleeland of The San Diego Union in December 1987. "It happens to everyone." He said that "you have to appreciate the potential for danger" but the record has "really been quite good for orcas - especially when you consider that about 40 people a year are killed in accidents with elephants." Stephens did not mention that many more elephants than orcas are kept in some form of captivity, and more people have close access to elephants because they are land animals. An orca cannot escape its pool and go on the rampage.

But is the kind of orca training programme important? What is a good programme? Sea World and certain other parks have long emphasized the number of behaviours you can train a whale to do. The splashier, acrobatic routines earn the most applause. Graeme Ellis, a former orca trainer at Sealand and the Vancouver Public Aquarium and an orca researcher who has collaborated on the pioneer study of wild orcas with Michael Bigg and others, says the measures of a good training programme are safety for the trainers, as well as keeping the orcas mentally healthy and interested.

Ellis believes the length of time a whale lives in captivity depends not only on the animal's age at capture and his personality but also on the trainer. "You have to be able to challenge them, to know how their minds work," he told me. "It's not how many tricks you can train them to do in two months; it's how long you can maintain a whale's sanity. They're curious enough and interested enough that they won't be driven neurotic in a year but it's difficult, because the novelty wears off. We seem to have a limited imagination when it comes to keeping these animals from becoming bored or neurotic."

Other trainers and marine park curators disagree, claiming that with good training, health care, food and companionship, most orcas can remain well-adjusted to captivity for life. At Sea World, the ability of some orcas to breed in recent years - six successful births at three of the Sea World parks in six years (1985-91) - is put forward as proof of their adjustment. However, as of January 1992, although six calves have survived, three of the four fathers and two of the five mothers have died. Two other females died following (but apparently unrelated to) miscarriages.

In the mid-1980s - about the time of the first births - David Butcher took over 'Supervision of all trainers at the Sea World parks, pushing a controversial new orca training programme. Butcher - with trainer Bruce Stephens who left Sea World in 1985 had already begun to emphasize human interaction as a reward for the whales, using an intentionally random system. Butcher pushed the programme further. Five types of training sessions known as PLESR (pronounced pleasure) were defined: play, in which anything goes; learning, when new behaviours are taught; exercise; socialization, in which animals interact with several trainers to simulate a pod; and relationship, when the animal and trainer spend time one-on-one to strengthen their bond. Sessions were scheduled at random, and Butcher kept a weekly log of all sessions between whales and trainers; if any patterns emerged, they were eliminated. If a whale misbehaved, Butcher and the other trainers retaliated by staring at it - a two-second slap of boredom, according to Butcher. Or did it communicate hostility? To further avoid predictability, no one animal could play the lead all the time; all got to play the role of the featured "Shamu".

It sounded like a promising idea to some, potentially dangerous to others. To its proponents, it was a way to alleviate the whales' and the trainers' boredom, to keep interest, as well as performance, high. This approach, soon dubbed the "Sea World Method," was adopted to some extent by other parks. But, along with the intended unpredictability of the training programme came a spate of accidents that eventually brought the whole programme into question. In March 1987, at Sea World in San Diego, Jonathan Smith, 21, was in the water performing with the orcas before several thousand cheering spectators crowded into Shamu Stadium. A six-ton orca suddenly grabbed him in its teeth, dove to the bottom of the tank, then carried him bleeding to the surface and spat him out. Smith gallantly waved to the crowd - which he attributed to his training as a Sea World performer - when a second orca slammed into him. He continued to pretend he was unhurt as the whales repeatedly dragged him 32 feet (9.8 m) to the bottom of the pool, as if trying to drown him. He was cut all around the torso, had a ruptured kidney and a six-inch (15-cm) laceration on his liver, yet he managed to escape and get out of the pool.

Did the whales have it in for him? Smith had spent only a few months with orcas, but had worked the seal and otter show for a year. Mild-mannered, he described the job as "learn-as-you-go." He must have wondered what he did wrong. Clad in his seal-like wetsuit, did he somehow trigger a hunting response? The whales' behaviour resembled typical seal-killing behaviour in some parts of the world. But there was no definite explanation.

Three months later, in June, Joanne Webber, 28 and a trainer for five years, had a three-ton orca, Kandu, land on her during rehearsal. She fractured a bone in her neck and has suffered permanent loss of head movement. In other incidents, Chris Barlow was rammed during a show and Mark McHugh was bitten on the hand while feeding a whale. That was only the beginning.

In August, the "accident" rate escalated. About a dozen accidents later, on November 21, 1987, Orky the mature five-ton male came crashing down on 26-year old John Sillick during a show in San Diego. At the time Sillick was riding on the back of a female orca. It was a crushing blow. Sillick almost died. He had severe fractures to both his hips, his pelvis, ribs and legs. After six operations in fourteen months, according to Sillick's lawyer, he was "reconstructed" with some three pounds (1.4 kg) of pins, plates and screws, including a permanent plate inserted in his pelvis and all his thoracic vertebrae permanently fused. He can walk today but his activity is limited.

After Sillick's injury, changes were finally made at Sea World. Sea World's owner, Harcourt Brace jovanovich (HBJ), the book publisher, stepped in with chairman William jovanovich calling the shots. The trainers were told to stop riding the whales, to stay out of the water with them, and to go back to the old training methods. Chief trainer Butcher was dismissed along with long-time zoological director and veterinarian Lanny H. Cornell and Sea World San Diego president Jan Schultz.

Various reasons are given for the accidents, but none can be fully explained. Some could have been simple miscalculations on the part of the whales, or missed signs. There might also have been poor signals from inexperienced trainers. The Orky incident with John Sillick did not surprise trainers who knew Orky, his history and his recent circumstances. Captured in April 1968, Orky had grown up with Corky at Marineland of the Pacific near Los Angeles, California. Soon after becoming mature, in 1978, he pinned trainer Jill Stratton on the bottom of the pool, nearly drowning her. "After that," said long-time Marineland head-trainer Tim Desmond, "we didn't regularly do water work with him because we didn't feel it was safe." On January 20-21, 1987, three weeks after Sea World owner HBJ bought Marineland, and promised not to move the orcas, Orky and Corky were trucked to San Diego, to join several other orcas. The new situation new breeding age females and much breeding activity - altered the social interactions of all the whales. But probably more important, Orky was being asked to perform according to the Sea World method with several trainers in the water. For Orky it had all happened almost overnight. Some Sea World trainers contend management was in too big a hurry to break in Orky. Did Orky have trouble making the transition? It is difficult to determine, but Robert K. Gault, Jr., then Sea World's president, admitted to the New York Times that they may have over-emphasized the importance of the entertainment, adding: "We did not have enough experienced trainers."

When Butcher had brought in his new methods and tried to standardize training in all four parks, he alienated several veteran trainers. In one year, about 35 trainers departed, according to Bud Krames, a senior trainer who left because he didn't agree with the system. New trainers had to be hired. Three of the five trainers in San Diego had three months or less experience working with orcas. Sillick, a veteran by comparison, had less than two years. In the year following the accidents, some of the injured trainers began to blame Sea World for not warning them about the "dangerous propensities of killer whales" as one lawsuit put it. Jonathan Smith's lawyer charged that Sea World and HBJ "negligently and carelessly owned, maintained, trained, inspected, controlled, supervised, located, transported and placed" the orcas, thereby exposing Smith to serious injury. Sillick and Weber also filed lawsuits. All three were later settled out of court with gag orders imposed. Following the terms of their deals with Sea World, the lawyers have refused to reveal any more than the basic details of their clients' cases. This means that no one can know any findings behind these cases; no one can learn or benefit from the thousands of pages of prepared evidence.

After Butcher left, Sea World revamped the training programme. Under new chief trainer Michael Scarpuzzi, they returned to a simple, consistent approach of rewarding each behaviour. Instead of going into the water, trainers began directing the whales from the deck with hand or underwater acoustic signals. But less than half a year later, they were back in the water. This time the animals were being taught to focus their attention on the main trainer on stage, ignoring trainers in the water. All of it was conveniently just in time to celebrate Sea World's 25th anniversary with a new orca show. The show must go on! Only time will tell if the current training programme proves safer than those of the past.

Since the first orcas were kept captive in the 1960s, there have been numerous "accidents" with trainers, most of which were covered up. Those that have come to light were mostly revealed by disenchanted trainers or members of the public who witnessed the accidents during a show. Marine park public affairs directors always played down such incidents, calling them bizarre accidents, and in some cases denied they had occurred. In recent years, with the proliferation of cheap video cameras, a number of incidents have been recorded. They range from bitings and collisions to near drownings when whales have held trainers underwater. Many of these dangerous incidents happened when the trainers were riding whales around the pool. Some former trainers such as Graeme Ellis believe that orcas, in general, do not like to be ridden. "They may tolerate it when they're young or new to captivity," says Ellis, "but later, it can lead to problems." Yet most marine parks still feature trainers riding orcas during the shows. Only Sealand and the Vancouver Public Aquarium in Canada, Miami Seaquarium in the USA, Marineland in France and Taiji in Japan no longer allow trainers to ride the whales. In recent years, fewer trainer accidents are known to have occurred at these establishments compared to parks that feature in-the-water work. Yet, there have been some injuries and the most serious incident of all occurred at Sealand.

On February 20, 1991, University of Victoria marine biology student and part-time trainer Keltie Byrne, 20, slipped and fell into the orca pool at Sealand of the Pacific. She had just finished a show with the three orcas. Since Sealand trainers stay out of the water, she was not wearing a wetsuit. One whale took her in its mouth and began dragging her around the pool, mostly underwater. A champion swimmer who had competed at the international level, she was no match for three huge orcas determined to keep her in the pool. At one point she reached the side and tried to climb out but, as horrified visitors watched from the sidelines, the whales pulled her screaming back into the pool.

"I just heard her scream my name," said trainer Karen McGee, 25, and then I saw she was in the pool with the whales. "I threw the life-ring out to her. She was trying to grab the ring, but the whale, basically, wouldn't let her. To them it was a play session, and she was in the water." McGee and other Sealand staff tried to distract the whales by throwing them fish, banging on the water with steel buckets and giving them hand and voice commands. Nothing worked. Byrne came up screaming one more time and then, as the whale swam round and round the pool with Byrne in its mouth, she finally drowned. It was several hours before her body could be recovered.

She had ten tooth marks on her body, the largest on her left thigh, but was otherwise untouched. The whales had stripped her clothes off. "It was just a tragic accident," Sealand manager Alejandro Bolz told newspaper reporters. "I just cannot explain it."

In May 1989, 21 months before the fatal incident, Sealand trainer Eric Walters told me he had quit the park because of differences of opinion over management. He felt Sealand was understaffed and unsafe. He complained to Sealand management before he left, but nothing was done. In April 1990, ten months before Keltie Byrne was killed, Walters wrote a detailed letter stating his concerns to the Canadian Federation of Humane Societies on the occasion of a national symposium on cetaceans in captivity. "Sealand of the Pacific is a dangerous place to work," he wrote. "I feel that sooner or later someone is going to get seriously hurt." His words were prophetic.

Walters cited problems with a training programme in which trainers were not allowed to enter the water with the whales. Since the whales were not used to having anyone in the pool, anything that fell in became very interesting. Yet the reaction of the Sealand whales in the Byrne case, Walters believes, may have been in part due to food and sensory deprivation.

Walters reports that some marine mammals including seals, sea lions and orcas were kept in a permanently "hungry" state at Sealand or deprived of food if they did not perform or co-operate. In an April 1991 letter to the British Columbia Veterinary Medical Association, Walters wrote: "If the killer whales did not enter the module pool [a small, dark, metal holding pool about 20 feet (6 m) deep and 26 feet (8 m) in diameter] at the end of the day to spend the night, we, as trainers, were instructed to withhold their end of-the-day allotted food. This was usually at least 25 to 35 percent of their daily food intake."

While in the module, the three whales, one male and two females, were barely able to turn around, much less escape from each other. They often cut or scratched their skin on the metal sides. Walters told me that he once saw the young male with flukes abraded and bleeding. As well, the orcas sometimes fought and suffered other injuries. Walters, now a biologist who has spent many hours observing wild orcas, said that the injuries were more severe than the usual rakes and scratches which result from orca play in the wild. On one occasion, the female, Nootka, was fighting with the others and crashed into the module, striking her head on the metal side. Her head was bleeding and blood came out of her blowhole.

According to Paul Spong, who has followed the case closely, the sensory deprivation imposed by 14 1/2 hours a day in tiny, dark quarters contributed to the mental state of the orcas that led to Byrne's death, as well as to several earlier incidents at Sealand. In one, trainer Henriette Huber was scratching Nootka's tongue when the whale bit her on the hand and pulled her into the pool. In another case, head trainer Steve Huxter was trying to retrieve a camera from a whale's mouth when the whale pulled him in, grabbed his leg and started to take him under. In both cases, Walters managed to rescue the trainers.

Food deprivation at marine parks is not generally discussed - even within the industry. Most marine parks build show times around feeding periods, and feed the animals after the show no matter how they perform. Food deprivation is officially considered antiquated by senior trainers, curators and other staff. It is generally accepted that a good trainer does not use, or need to use, deprivation. As well, some such as Sonny Allen, director of marine mammal training at Marine World Africa USA, have advised against deprivation because it "can work adversely down the road." He told the 1989 International Marine Animal Trainers Association annual conference, "what happens, we've found, is that it leads to aggression."

Yet certain trainers do at times withhold some food if an animal refuses to perform. A former Sea World trainer who requested anonymity told me that whales or dolphins that would not perform were sometimes denied food during or immediately after the shows. They would only be given their "base" including vitamins - about 2/3 of their daily food allotment. "Usually the whales would start performing when they realized they weren't going to get fed," she added.

But Paul Nachtigall of the US Naval Ocean Systems Center, found that food deprivation did not enhance learning in bottlenose dolphins, and that hungry animals were even less interested in co-operating.

Some of these ideas and theories were discussed at a public inquest in the Byrne case, in Victoria, in May 1991. Sealand was criticized for poor safety and emergency features and for the practice of confining the whales at night in the holding tank. But there was no outright criticism of food deprivation. The inquest resulted in a list of twenty recommendations for preventing a reccurrence of tragedy. Sealand was told to refrain from keeping whales penned up in the holding tank unless it was necessary for veterinary or husbandry purposes. (Part of the reason for locking them up was that Sealand has only a gate separating the orcas from the open sea, and Sealand's owner apparently feared the whales would escape or be let go.) The jury recommended that Canada's Department of Fisheries and Oceans develop a marine mammal policy to ensure the health and welfare of captive whales and to make scientifically-based inspections and assessments of each aquarium twice a year. To date, no government action has been taken. But, following Byrne's death and fearing that they would be closed down permanently, Sealand management began instituting safety measures such as installing railings around the pool and placing air tanks and scuba gear nearby. Head trainer Steve Huxter started using a long handled brush and roller to give the whales more contact. As well, Huxter worked with the three orcas on "desensitization training" so the whales would ignore a new object in the pool. There was no plan to train the whales to accept trainers in the water, to institute an in-the-water programme - the suggestion of some trainers.

But it was all too late for Keltie Byrne. Her parents have decided so far not to sue Sealand, preferring to put the tragedy behind them. The jury at the public inquest was unable to agree on the real cause of Byrne's death, beyond drowning. Why did orcas, which had never killed a trainer in marine parks or in the wild despite thousands of encounters, suddenly kill a human? Was it "an accident waiting to happen," if not at Sealand, then at Sea World or almost any park, especially one where basic safety procedures are overlooked? Was it the inevitable result of keeping orcas in captivity, a situation in which behaviour is shaped by young human trainers and influenced, even distorted, by the physical and social conditions imposed by life in a small enclosure, with the day-in day-out demand to perform and to live in close proximity with humans and animals that they would never socialize with in the wild? Or, was the bizarre behaviour the specific result of the routinely imposed sensory deprivation when the Sealand orcas were confined in the tiny metal module?

The jury did not stop Sealand's orca shows, but the city in which Sealand is based has indicated that Sealand's lease may not be renewed. In September 1991, Sealand owner Bob Wright put the three orcas up for sale. But what marine park wants to take three orcas that killed their trainer? Even before Sealand announced the whales were for sale, Sea World was preparing an application to NMFS to import them.