WASHINGTON - The engineer on Amtrak Train 188 remembers speeding up, he remembers a curve in Philadelphia that can "sneak up," and he remembers his body lurching as the locomotive entered the bend much too fast just before it derailed last May, killing eight people and injuring more than 200.

But the engineer, Brandon Bostian, told investigators that he remembered little about the exact moment the train hurtled off the rails. And he did not explain why it sped to 106 mph as it approached a 50-mph curve, according to transcripts of two interviews released by the National Transportation Safety Board on Monday.

When Bostian came to, emergency crews already were pouring into the scene of one of the most devastating rail accidents in recent memory.

Bostian's statements were part of more than 2,000 pages of documents the safety board released as it completed the fact-finding phase of its investigation and turned to its analysis of what went wrong.

Included are harrowing accounts of crew members thrown through the train; police calling for more tourniquets; and one conductor groping for his hat amid the wreckage, "so that people would be able to see someone in charge of the situation."

Investigators' summaries also detail the trip's final moments as the train traveled north. Bostian had pushed it to full throttle as it approached the Frankford Junction curve, hitting the emergency brake just three seconds before the train derailed.

The documents confirm much of what the NTSB has already said. There were no mechanical, track, or signal problems. There was no "smoking gun," an NTSB official told reporters.

Bostian, 32, was not under the influence of alcohol or drugs, and was not using his Samsung Galaxy cellphone for calls, texts, or data. He told investigators he was in fine health.

NTSB investigators will use the documents as they try to determine a probable cause for the crash and make recommendations to improve rail safety. Those steps are expected in the spring.

But with so many factors ruled out, much attention has focused on human error and why Bostian sped up as he approached the curve.

Some engineers and members of Congress have speculated that he simply lost awareness of where he was on the trip from Washington to New York. Along that route, on some stretches, the speed limit increases to 110 mph.

Bostian's interviews - one three days after the crash, another six months later, in November - are his most extensive public comments so far.

Praised as attentive by fellow crew members, he told investigators that he didn't look for speed signs - "a lot of times they're either missing" or wrong - and that it would be easy to enter the curve "a little bit hot" because the start was hard to discern.

He had worked trains in the area of the crash for three years, but had been an engineer on this route for only about two weeks. The night of the crash was the last of his five-day work week. He told investigators he was not tired.

Bostian, like all Amtrak engineers, was tasked with memorizing his route, and he recited for investigators the changing speed limits along the way and the visual markers he'd seek out near the curve, such as an elevated train bridge and a signal box that "sometimes you can see it and sometimes you can't."

On this night, the sign was in place and properly reported a speed limit of 50 mph.

There was confusion several minutes before the crash: an object, perhaps a rock, had shattered the window of a SEPTA train, and Bostian worried that crew members might be on the tracks.

He also realized he was going 70 mph when the speed limit was 80 mph, and sped up, he told investigators.

"And I don't remember anything from that point until after the train was already in the curve," he said.

In the curve, he felt his body lurch.

"That's when I realized that it wasn't that the train was going somewhat fast around the curve," he said. "The train was going significantly fast around the curve."

He threw on the emergency brake, "feeling like, OK, well, this is it, I'm going over."

He came to his senses standing in the cab after the crash.

Bostian had pushed to full throttle - "notch 8" - about 55 seconds before reaching the curve, the train's data recorder showed. He held it there for 30 seconds, slowed to notch 7 for two seconds, then pushed back to full throttle for 20 seconds more.

He hit the emergency brake - too late.

Bostian said his memories were foggy, and at times said he wasn't sure if they were from the night of the crash. Some, he said, may have been confused with other trips.

He said it was easy to enter the curve too fast "if you weren't being careful, and looking very carefully at the cues, because it can sneak up on you."

"You don't see the curve until you're in the curve, basically," he said.

While Bostian told investigators he could not remember the moments before or after the crash, NTSB officials noted that he did recall radio messages between a SEPTA train and a dispatcher that occurred after he passed the North Philadelphia Station.

And lawyers for nearly 30 victims accused Bostian of changing his story between the two interviews, seeming to recall more detail months later.

Bostian's lawyer did not respond to a request for comment Monday.

A few weeks after the crash, Bostian's train engineer's certificate was suspended, the NTSB documents show.

At the same time, a fellow conductor told investigators Bostian was "on top of his game. Knowledgeable. Always knew the answers. Very knowledgeable of the territory."

Another said Bostian "knows his job. He's there on time for the briefing. He answers any questions we have. . . . I've never seen him do anything that he wasn't supposed to do."

Bostian did not speak to Philadelphia police, but an NTSB official said he was "extremely cooperative" with the board.

Crew members recalled violent shaking as the train entered the curve.

"It seemed like forever," said a conductor, Akida Henry. "We rolled and slid and the train was on its side. After - you could just see stuff blowing up. "

The carnage became clear almost immediately.

"We have people on the track. We have a major incident with several cars rolled over," one responder said early on, according to emergency radio transcripts. A Fire Department official cut in: "We're going to need multiple, multiple medical units."

Two people were pinned beneath the train, dead.

Police and fire officials told NTSB officials of a chaotic scene: pitch-black darkness filled with hundreds of responders, walking wounded, bodies in open view.

Anthony J. Sneidar Jr., a deputy chief in the Philadelphia Fire Department, described the scene as "Dawn of the Dead: bodies and people and things broken and bleeding and cops and firemen and everybody. So my thought there was, all right, this is crazy."

Thomas O'Brien, a conductor, was winded but unharmed. His first move was to put his hat back on.

"They're going to want to see a guy that looks like he knows what he's doing and looks like he's not ready to wet his pants and cry," he told investigators.

Rocks poured on his face, O'Brien remembered, but he tried to help people evacuate.

The first woman he saw, "I couldn't see her eyes anymore. . . . There was just all blood and she's screaming about she can't see."

Pictures included in the NTSB report show seats wrenched free from their frames, wires hanging loose in the aisles, and train interiors coated in dirt, tree limbs, and debris. Investigators found blood streaked throughout.

The two passenger cars closest to the locomotive were so damaged, investigators could only survey their exterior. Yet in the café car, along with a hole torn in the wall, a coffeemaker stayed in place.

Joseph Brennan, a dispatcher who was riding the train, said he borrowed a cellphone from a fellow Amtrak worker.

"His face was beaten up pretty good," Brennan told investigators.

It turned out to be Bostian. Brennan asked what happened.

"And he just looked at me and he said, 'I don't know.' "