WASHINGTON — For nearly three years, the military held the key to Roger House's exoneration and didn't tell him: A forensics examiner had botched a crucial lab test used in the Navy lieutenant's court-martial.
In fact, the military had begun second-guessing a decade's worth of tests conducted by its one-time star lab analyst, Phillip Mills.
Investigators discovered that Mills had cut corners and even falsified reports in one case. He found DNA where it didn't exist, and failed to find it where it did. His mistakes may have let the guilty go free while the innocent, such as House, were convicted.
"It cost him his family and it cost him his Navy career," House's attorney, John Wells, said in an interview. "It's certainly outrageous and unconscionable; it's the kind of action that makes you want to scream."
But the problem was bigger than just a lone analyst.
While a McClatchy investigation revealed that Mills' mistakes undermined hundreds of criminal cases brought against military personnel, it also found that the U.S. Army Criminal Investigation Laboratory, near Atlanta, was lax in supervising Mills, slow to re-examine his work and slipshod about informing defendants. Officials appeared intent on containing the scandal that threatened to discredit the military's most important forensics facility, which handles more than 3,000 criminal cases a year.
The military has never publicly acknowledged the extent of Mills' mistakes nor the lab's culpability. McClatchy pieced together the untold story by conducting dozens of interviews and reviewing internal investigations, transcripts and other documents.
The McClatchy investigation shows:
Taken together, the mistakes shocked veteran military officials.
"My confidence in your entire lab is shattered," Navy Capt. Bruce MacKenzie, a military judge, told lab officials in 2008.
The lab's reputation matters because DNA and other forensics evidence are crucial even beyond solving crimes and exonerating suspects. Forensics evidence can embolden prosecutors to indict, and frighten defendants into confessing. The military's own inquiries have called all this into question.
"The result," prominent Penn State forensics analyst Robert Shaler summed up, after an independent review of the lab's mistakes, "might have led to a miscarriage of justice."
The U.S. Army Criminal Investigation Command, which oversees the lab's 180 employees, declined to grant a sit-down interview or to provide Mills-related documents.
In an extensive written statement, the Army defended its actions and maintained that Mills' conduct didn't reveal systematic problems at the U.S. Army Criminal Investigation Laboratory.
"It is important to point out that this was never a 'USACIL integrity' question per se, but an issue with an employee's integrity," the statement said.
The Army said its $1.4 million retesting of Mills' evidence demonstrated that the lab took the revelations seriously.
"Although this should never have happened, it did, and we feel very strongly that we took immediate corrective action and have done everything possible to prevent this from happening in the future," the statement said.
Mills couldn't be reached to comment, either by phone or mail, over several weeks. But in the past, he's defended his work.
"I've always followed the proper procedure in making sure that I do the same thing over and over in each case," he said at a 2008 court hearing.
SIGNS OF TROUBLE
For almost 30 years, the military trusted Mills and took him at his word, but lab officials overlooked danger signs that might have alerted them to a looming crisis.
In 2002, Mills failed a hair-analysis proficiency test. In 2003, he mixed up DNA samples. By then, he'd begun to get a reputation for sloppiness among those who signed off on the lab analysts' work.
"Too often the review of Mills' cases takes an extraordinary amount of time, and the reviewing examiner is ultimately performing some of the work Mills should have," lab biologist Thomas Overson wrote in a 2005 memo.
Overson reviewed Mills' work after the 2003 incident and thought it suggested a pattern of mistakes. He recommended a broader review. Mills' supervisor, Clement Smetana, suspended Mills, but declined to open a wider inquiry. Smetana and other lab officials saw Mills as one of their most productive and experienced analysts, which made them reluctant to confront him.
In another lapse, lab officials didn't tell defense attorneys. They reasoned that since the botched DNA test was caught, public notice wasn't necessary.
Since the military didn't preserve evidence, it also missed a chance to retest work that Mills might have gotten wrong.
Ivor Luke, for one, has been thwarted in his effort to challenge his 1999 conviction because of the destruction of evidence. An enlisted woman had accused the Navy hospital corpsman of sexually assaulting her aboard the USS Port Royal while he was conducting a medical exam.
"The government ... destroyed the physical evidence ... thereby precluding the type of retesting that might have restored some level of confidence in the process," Chief Judge Andrew Effron, of the U.S. Court of Appeals for the Armed Forces, noted earlier this year.
Despite the seriousness of the 2003 mix-up, Mills returned to casework the next year. Soon, his work was flagged again when co-worker Dr. Timothy Kalafut found that Mills falsified DNA paperwork, and Kalafut alerted supervisors.
Kalafut and other lab employees grew impatient with the pace of the lab's response, however. "We felt like it was taking an unacceptably long period of time to even confirm if a problem was found," Kalafut said.
Though they seemed slow to react, lab officials were deeply alarmed by Mills' errors.
The Army criminal investigation lab's reputation was at stake. Officials feared that the lab's accreditation might be revoked. They worried that they might have to withdraw results from a forensics database.
Lab officials could have authorized an independent review. Instead, they decided that the lab would investigate itself.
"If perception is truly the paramount concern (and confirming Mills' results is essential), then USACIL needs to retain control of retesting decisions in each case as they come up," lab attorney Lisa Kreeger said in a 2005 e-mail.
Kreeger also floated the idea of hiring a former judge with DNA expertise to monitor the review, saying in an e-mail that the jurist would "make one hell of a government witness" and could "formidably pre-empt any unreliability argument" if defense attorneys challenged convictions.
Internal documents show that officials hoped to retain control of the inquiry, even as they hired Shaler, the Penn State expert. Shaler, who later proved very critical of the lab, was prohibited from speaking publicly about his work.
The lab originally envisioned a six-month review, but it grew into what one lab official called a "three-year research project on a worldwide scope."
One reason is that some investigators balked at reopening old cases.
Military higher-ups were reticent, too. They struggled with their legal obligation to share evidence that had the potential to clear defendants. Rather than individually notifying every defendant implicated by Mills' testimony, they took a narrow approach.
The lab sent a general notice to each military branch's legal division in August 2005, reasoning that individual notices were required only if all the evidence was to be destroyed by retesting.
The limited notification meant that many never learned of the retesting. Attorney Charles Gittins, for instance, didn't hear about evidence discrepancies involving his client Troy Jenkins, an enlisted man who pleaded guilty to charges of rape. Jenkins has claimed he felt pressure to admit his guilt because of the DNA. His co-defendant later went to trial and was convicted on lesser charges.
"I would have liked to have been told," Gittins said in an interview. "Isn't that something that you send a certified letter for? It raises questions about my client's guilty plea."
Others heard over the prison grapevine.
Former Marine Corps Gunnery Sgt. Steven Carlson was serving a 15-year sentence for charges related to sexual assault when he learned of the Mills debacle, but it wasn't the military that told him.
Carlson said another inmate informed him in September 2005 that there was a lab investigation and it involved Carlson's case. He took matters into his own hands.
"I paid for the memo, and made photocopies of it," Carlson said, adding, "Nothing in prison is free."
A CACHE OF LOST EVIDENCE
In interviews, lawyers said they were troubled that many defendants weren't told of the lab's review. Military law generally allows only two years to petition for retrial based on new evidence, less time than in civilian courts.
It wasn't just the defense that was undermined by Mills' errors. Prosecutors might have missed the chance to detect the guilty. The lab review found 49 cases in which Mills' testing didn't identify stains or DNA. Investigators called these "questionable," because of his lack of thoroughness.
"It's absolutely possible that someone slipped through the cracks," said one military attorney who asked to remain anonymous because he might be disciplined for discussing the matter.
"The military should have said: 'Stop the presses. We've got a really big problem here,' " this attorney added. "They surely could have done more than just send out a memo."
Lab officials feared the fallout, though. Behind the scenes, documents show, they crafted talking points. Officials were instructed to respond to certain questions only if pressed. Lab officials spoke of ways to "mitigate some of the negative news."
"The military just wanted to make the Mills scandal go away as quickly and conveniently as possible," said attorney Peter Griesch, who formerly represented Luke.
But as lab officials were preparing to defend themselves, they were about to be undermined from within.
In April 2008, three years after Mills' falsified report, lab employees found a trove of lost evidence. In Mills' old locker, they found four containers of slides. Inside a closet, they found more slides. And in a refrigerator, they discovered evidence from 36 Mills cases.
Over many months — perhaps even years — lab supervisor Smetana had known about the material but never turned it over to investigators. Smetana, a lab employee of 30 years, resigned in July 2008, shortly after investigators found him "derelict" in his duty.
All the newly discovered evidence had to be retested, delaying the investigation.
In September 2008, the lab completed its review, and it was withering. Mills often "did not follow good scientific practice." His paperwork documenting his analyses was "lacking." He used up too much evidence, making it impossible to retest. In nearly one-quarter of the cases, reviewers identified problems.
Not every disagreement meant Mills got it wrong. Some evidence could have degraded over time. Technology has become more refined.
Still, Mills' errors cast doubt on every DNA profile he submitted to a national database. Lab officials had to withdraw 67 suspect profiles, complicating police work.
Shaler, the outside reviewer from Penn State, was even more critical, government documents show. He attributed Mills' errors, in part, to "an innate intellectual dishonesty." Shaler also blasted lab management. He warned that the lab lacked proper safeguards and permitted analysts to invent laboratory procedures that allowed mistakes to go unnoticed.
In its statement, the Army defended its handling of Mills' mistakes, disputed some of Shaler's criticisms and says it "took the situation very seriously." The lab has made changes, including replacing managers and imposing stricter case reviews.
The lab "has invested time, money, training and personnel to improve overall capabilities," the Army said, adding, "We have made more than 100 extensive changes and modifications."
But defense attorneys whom McClatchy interviewed said that the Army should have ensured that Mills was held fully accountable.
The lab committed to providing its final review to a federal prosecutor, David Leta, so he could determine whether Mills' conduct violated federal law, including the prohibition against making false statements to federal officials. Lab officials say they mailed the report in 2008. Leta, though, says he heard of it only in February, when McClatchy informed him.
He's now reviewing the case.
Some defendants implicated by Mills, meanwhile, are trying to seek justice on their own.
House, the Navy officer who was wrongfully convicted, filed a federal lawsuit seeking back pay and a promotion he was denied. Luke, the enlisted man whose evidence was destroyed, is preparing an appeal to the U.S. Supreme Court.
Others, such as former Gunnery Sgt. Carlson, appear close to giving up hope.
"I always thought the truth would prevail," Carlson said, "but after almost 12 years, my faith has dwindled."
(McClatchy researcher Tish Wells contributed to this report.)
MORE FROM MCCLATCHY