On a Connecticut summer evening 45 years ago, Edward May was operating a northbound commuter train in Darien when the lights of another train suddenly appeared on the tracks, 600 feet away and closing fast.
It was Penn Central commuter train N-48. Its engineer, running late, disregarded a message to yield to May's train.
"He just kept coming at me," May, 82, of Waterford, recently recalled.
The two trains, each traveling 30 mph, collided with such force that N-48 telescoped halfway through May's locomotive. Three crew members and one passenger died and 43 passengers were injured.
Rescuers spent five hours cutting May from the wreckage.
The engineer who struck May's train had known about the order to pull onto a side track; a copy of it was found on his body. Investigators never learned why he failed to yield.
By rail-disaster standards, the Aug. 20, 1969 crash on the New Canaan Line wasn't the most tragic or horrific of the era. But it stands as a milepost in rail safety history: the beginning of the National Transportation Safety Board's drive to install technology to remove human error from railroad operation.
After investigating the Darien wreck, the NTSB issued its first formal recommendation to the Federal Railroad Administration to "study the feasibility of requiring a form of automatic train control at points where passenger trains are required to meet other trains."
Positive Train Control, as the technology would become known, monitors and controls train movements with a digital communications network that links locomotives with control centers. It is designed to prevent collisions by automatically slowing or even stopping trains that go too fast, miss a stop signal, enter a zone with maintenance workers on the track, or encounter other dangers.
Indisputably, PTC saves lives. "The equivalent of air traffic controllers," U.S. Sen. Richard Blumenthal, D-Conn, said of PTC.
A Hearst Newspapers investigation found that the NTSB's push for PTC encountered a long and twisting road.
The rail industry resisted implementing the system to preserve profits and the FRA went along, backing the industry it regulated by arguing against the safety standard and ignoring repeated NTSB recommendations to mandate its implementation, the Hearst review found.
Over the decades, the NTSB classified 139 crashes on America's rails as PTC-preventable. Those wrecks resulted in 288 deaths, 6,500 injuries and nearly $300 million in property damages, federal safety records show.
Ridgefield resident and passenger Jay Eldon Paris, 55, died in the 1969 Darien collision, which also took the lives of Frank J. Bojarski, 43, of Stamford, the engineer of the northbound train, and Costas Vlahakis, also of Stamford, a conductor on the southbound train. Eugene A. Chain, 34, of West Haven, a flagman on the southbound train, was also killed.
Adding to the list of rail accidents was the Metro-North derailment last December in the Bronx that claimed four lives and injured 75 passengers after an inattentive engineer barreled through a corner at twice the posted speed.
Congress in 2008 finally mandated PTC be installed on the nation's trains by December 2015, but it's unlikely the deadline will be met and efforts are underway to extend it to 2020 and beyond.
Today only a tiny slice of America's rail network has the safety system -- portions of Amtrak's northeast corridor and on its Michigan tracks, and a section of the Metrolink passenger rail system in Los Angeles.
Cash-strapped commuter railroads have been victimized by broken promises from Congress. Metro-North Railroad and other passenger lines were promised a $250 million appropriation to help with PTC installation in 2008, but only one-fifth of that money came through.
The human factor
Human error is the cause of roughly 40 percent of train derailments, federal records show.
One of the worst happened in 1972 when the engineer of an Illinois Central Gulf commuter train backed up just enough after missing a station to give the train behind it a green light to proceed. The speeding train knifed into a double-decker car at the rear of the backing train, killing 45 and injuring 332.
After the crash, the NTSB challenged the Urban Mass Transportation Administration, and later its successor, the Federal Railroad Administration, "to justify or disprove" the need for a train control system.
That challenge was just the latest in a decades-long series of recommendations to adopt PTC.
A Hearst study of accident reports since 1969 shows the NTSB issued 75 PTC-related recommendations -- formal advice to the government and railroad industry -- and those recommendations took on increasing urgency.
The FRA responded defiantly to the NTSB's recommendations, a review of correspondence shows.
The frequency of accidents due to incapacitated engineers is "so very small that a requirement to install an expensive hardware system on the entire locomotive fleet is not required or necessary," the FRA wrote.
The NTSB declared the railroad agency's response "unacceptable," noting it was investigating still more accidents involving human error. The NTSB grew "frustrated," as the safety agency put it, at what members viewed as an effort to mischaracterize train control and inflate its cost.
"The FRA has overestimated the cost of (PTC) installation by taking the most expensive figure available and extrapolating it nationally to every single mile of track," according to a 1987 letter by the safety board.
After a 1993 PTC-preventable crash killed five in Washington state, the NTSB again turned up the heat, advising the FRA to establish a "firm timetable" for installing train control along America's railroad tracks.
But the opposite occurred and it would take several more decades and Congressional action to force implementation of PTC.
Grady Cothen, an FRA deputy associate administrator until he retired in 2010, said "The whole issue around us was that it was all about costs and benefits from the view of industry."
A PTC prototype
During the 1980s, railroads began looking at train control systems as a potential business benefit: trains could run faster if technology better regulated their spacing on tracks. U.S. railroads collaborated with Canadian companies in a research effort called Advanced Train Control Systems.
Burlington Northern, plagued by a series of fatal accidents, was the first to take significant steps to address an already two-decade-old NTSB safety recommendation to adopt PTC.
The railroad's Advanced Railroad Electronic System, which resembles modern PTC, plotted speed and real-time positions of trains within 30 feet. If an engineer didn't slow after warnings flashed on a locomotive computer screen, the system took over the train. For five years, Burlington Northern's system operated 17 locomotives on 300 miles of tracks on northern Minnesota's Iron Range while using the system.
"All of the components worked as expected," said Steven Ditmeyer, who was Burlington Northern's research director at the time, in a recent interview.
"We had acceptance by train crews, dispatchers and maintenance people. There was no fear of the system and people could see its benefits," Ditmeyer said.
Mergers and shifts
In 1993, the Association of American Railroads prepared a confidential study, obtained by Hearst Newspapers, which laid out a robust case for benefits of PTC technology beyond avoiding wrecks: savings in fuel and labor costs; better traffic control; a means to monitor the health of locomotives; and "a better-rested and safer workforce."
"With costs declining and benefits probably understated, it is time for the railroad industry in North America to make an investment in the future," the report asserted.
But rather than use the findings to rally its members, leaders of the railroad trade group ordered the study destroyed.
Charles Dettmann, the American Association of Railroads' former executive vice president for safety and operations, recalled that business benefits "were not concrete enough in the '90s, which made it very, very questionable about committing that much capital."
"Nobody was offering any money. It was all on our own hook," Dettmann said.
Ditmeyer, who headed the FRA's Office of Research and Development after being involved in the successful Burlington Northern PTC project, testified during a 1996 congressional hearing that technical issues still needed to be addressed.
In a recent interview, Ditmeyer recalled the testimony as "one of the things I regret most in my life. ... I was forced to say it was not ready to implement."
The 1990s was a period of upheaval in the industry, with mergers set in motion by deregulation. The freight rail industry shrank from more than two dozen in the early 1980s to the seven Class I railroads of today, and soon lost interest in train control.
Frank Wilner, assistant vice president of the Association of American Railroads during that period and author of the book "Railroad Mergers," said money and energy "were siphoned off" during the merger wave.
"Attitudes shifted and the whole project was shelved," he said.
The new "super railroads," expressing fear the government would force PTC on them, refused to even consider potential business benefits when costs and benefits of the technology were assessed.
George Gavalla, who headed the FRA's safety office from 1997 to 2004, said his agency needed political support from Congress to mandate the life-saving equipment.
"The industry has a very good sense of what is politically feasible and what is not. They are big players," he said.
Former U.S. Rep. James Oberstar, who watched the potential of PTC on the Iron Range in his Minnesota congressional district, used his leadership position on the powerful House Transportation and Infrastructure Committee to make PTC the heart of rail safety legislation in the 1990s.
"They (the industry) bottled that bill up for 12 years," said Oberstar, who died in May, in one of his last interviews. "They had a very strong lobbying effort. Railroads are very effective on Capitol Hill."
Indeed, the industry has displayed legendary clout over the years.
Last year alone, the Association of American Railroads and the seven biggest railroad companies spent at least $35 million on Washington lobbying, a review of lobbying records shows.
The list of lobbyists railroads hired reads like a who's who in the Washington influence machine: former senators, John Breaux of Louisiana and Trent Lott of Mississippi; at least a half-dozen former U.S. House members; former National Republican Chairman Haley Barbour; and Jeff Loveng, who was chief of staff to Rep. Bill Shuster, R-Pa., chairman of the House Transportation and Infrastructure Committee.
Favoritism or not, the FRA was channeling railroads' concerns about the cost of PTC.
In a 2004 letter to Congress accompanying a cost-benefit study, Bette Monro, a deputy FRA administrator who headed the railroad agency in an acting capacity, wrote that PTC-preventable crashes amount to just 2 percent of train wrecks and "the direct safety benefits that would accrue to railroads are small relative to the costs."
Monro resigned her post after the Transportation Department's inspector general issued a report on her friendship with a Union Pacific lobbyist, which included sharing a vacation house.
Inside the FRA, PTC advocates grew frustrated at the lack of progress.
"We knew it would take a horrible accident for people to say enough is enough, and that Congress would have to make the judgment about PTC," said Grady Cothen, the FRA's former safety chief.
In fact, it would take two more horrible accidents to prompt action on PTC: a 2005 Graniteville, S.C., accident that claimed nine passengers and a Chatsworth, Calif. crash in 2008 that killed 25.
By the time the California disaster occurred, the House and Senate each had passed rail safety bills with PTC mandates. Within weeks, the differences were ironed out and President George W. Bush signed Positive Train Control into law.
Robert Sumwalt, an NTSB board member, said he and other members "were feeling good" after Congress ordered PTC six years ago, in 2008.
"And now we're finding that it's going to be delayed even further," Sumwalt said, referring to a GOP House plan to delay implementation. "It's frustrating to see accidents continue to happen that we know PTC would have prevented," Sumwalt said.
Bill Lambrecht is an investigative reporter for Hearst Newspapers.