FBI bullet test misses target Court rejects test; FBI has suspended use. By Leonard Post Staff reporter Monday, April 4, 2005 Hundreds of criminal cases nationwide could follow the course set by a recent New Jersey appellate court that reversed a murder conviction after rejecting the validity of a test used by the FBI to analyze bullet lead. Just last week, the FBI disclosed that it had suspended the use of the test-called comparative bullet lead analysis (CBLA)-since the release of a February 2004 National Research Council (NRC) report that severely criticized the FBI for overstating the conclusions that could be drawn from the tests. But FBI Special Agent Ann Todd noted last week that FBI officials will continue to testify in cases where the analysis has already been done. Todd said that the FBI is "incorporating the NRC recommendations in our testimony." She said that the lab had compared two possible options recommended by the NRC against its prior methodologies and selected one-which she could not yet disclose-for potential future use. The New Jersey Appellate Division ruling, New Jersey v. Behn, No. A-2062-03T3 (N.J. App. Div.), gives new imperative to an ongoing Freedom of Information Act (FOIA) suit, filed in 2004 against the FBI in an attempt to identify all the defendants against whom testimony involving the test was used. The plaintiff in the FOIA suit says that many prisoners may not be aware of the growing chorus of scientists who have challenged the legitimacy of this testimony, which could have caused or contributed to their convictions. The FBI so far has said that, since 1986, five agents have testified 521 times about CBLA tests, but the bureau does not say where or against whom. In its March 25 pleadings in the FOIA suit, it states that the information is not readily available. Forensic Justice Project v. FBI, No. 1:04CV01415 (D.D.C.). "If they don't know, they have a huge quality-assurance problem," asserted the plaintiff's attorney in the FOIA suit, David K. Colapinto of Washington's Kohn, Kohn & Colapinto. Behn, the New Jersey ruling, foreshadows trouble for prosecutions that were won on the back of CBLA, said Howard Weiner, a criminal defense attorney in New York's Law Offices of Lawrence S. Goldman. That is why the FBI is resisting the FOIA suit, he said. "It's not just the trials, but all the defendants who plead out because for decades there was no way to challenge CBLA testimony," said Weiner. "Now that we know it's junk science-that's new evidence-and we need to find a way to spread the word to prisoners, some of whom are surely on death row." First in nation In the New Jersey ruling, which stemmed from a habeas petition that was denied in a lower court, convicted murderer Michael Behn had asked a New Jersey appellate court to grant an evidentiary hearing. He alleged that the FBI's scientific testimony was unfounded, and thus he was entitled to a new trial. Deciding that a hearing was unnecessary, the appellate court last month unanimously overturned Behn's conviction. It said that the "integrity of the criminal justice system is ill-served by allowing a conviction based on evidence of this quality, whether described as false, unproven or unreliable." The Behn court was the first in the nation to overturn a conviction based on the statistical assumptions that FBI witnesses have drawn from CBLA. The test has been used in thousands of cases in which a person is shot, but no weapon is found, or in which the crime scene bullet is too damaged to match to a weapon. Crime scene bullet lead or bullet lead fragments are tested for the presence, quality and quantity of other elements-such as cadmium and copper. Next, that same analysis is done to bullets that somehow have been tied to a suspect. Then both are compared and a determination is made as to whether they are analytically indistinguishable, and ultimately whether to infer that the samples came from the same manufacturer, or the same batch made from the same molten source, or from the same box of bullets. Charles Peters, an FBI specialist in bullet-lead analysis, had testified that the fragments recovered from the victim and Behn's bullets came from the same box or boxes that were packaged by the manufacturer on the same date. This was just one of Peters' conclusions that the appeals court found unsustainable. But the court had help. To Behn's rescue In early 1999, William Tobin, the FBI's chief metallurgist who had recently retired from the FBI after 24 years, was contacted by Behn's sister, who asked him to review Peters' testimony and his laboratory data. Tobin was troubled by what he saw, and thought the problems were endemic, not case specific. He asked Erik Randich, a metallurgist at the Lawrence Livermore National Laboratories in Livermore, Calif., to collaborate. They researched the scientific validity of CBLA for about three years. The scientists said in Behn's petition that it was false to conclude that any particular lot of lead supplied to a manufacturer was unique. And they claimed that Peters, a scientific examiner, was incorrect when he testified that Behn's bullets and those at the crime scene were manufactured on the same day. In fact, Randich told the court that he had found compositionally indistinguishable bullets manufactured 10 years apart. Tobin told the court that the bureau's lab was the only one in the country doing CBLA, and that until 1993 that analysis required the use of a nuclear reactor, which no civilian had access to. He said that no civilians had studied CBLA prior to 1998, when Tobin began to look at the science. The FBI laboratory database was not available to civilian researchers until 2000, he said. Tobin found that database flawed. In 2000, the court noted that the FBI had commissioned a study by the department of statistics at Iowa State University to decide whether it was possible to determine the statistical probability that two bullets came from the same source. That study concluded that the FBI did not have a sufficient database to reach such conclusions. "The research has now shown that there is no basis for the claims that the FBI has been making over the years," said solo practitioner Paul Casteleiro of Hoboken, N.J., Behn's post-conviction attorney. "The courts are no longer going to let them get away with claims based on outlandish facts. The tide has swung." Simon Rosenbach, a Middlesex County, N.J., assistant prosecutor, said his office would seek New Jersey Supreme Court review of Behn by writ of certiorari. "You bet ya [we're filing]," Rosenbach said. "It was inappropriate for the appeals court-which is not a fact-finding court-to decide the case based on two unchallenged affidavits and two unsworn expert reports. At worst, they should have granted a hearing." The FBI declined to comment on the case. More ammunition In 2004, the National Research Council (NRC) of the National Academy of Sciences issued a report on CBLA after being asked by the FBI to study its scientific validity under guidelines it provided. The report came out after the lower court's denial of post-conviction relief in Behn, and too late for the appeals court to consider. The NRC report-"Forensic Analysis: Weighing Bullet Lead Evidence"-pointed to possible bias in the FBI's database and said that the bureau should change its statistical analysis for the technique. Also, the report recommended that due to variations in the manufacturing process, the FBI should make clear the very limited conclusions that CBLA results can support. In other words, "Short of physically following the bullets, it is never possible to conclude that bullets come from the same batch or the same box based on CBLA," Tobin said. A Feb. 10, 2004, press release from the FBI said that the report had approved of the "precision and accuracy" of the FBI's measurements, and that the NRC had "included suggestions to improve the statistical analysis." It added that the "science has continually withstood legal challenges in federal, state and local criminal courts" in the 2,500 cases since the early 1980s in which it has been used. Less than 20% of those cases involved presenting circumstantial evidence in court, it said. Doubt over CBLA continues to trigger court action. In February, in Iowa, Albert Ware, who was convicted of murder in 1982, filed an application for post-conviction relief. He claims that an FBI agent had tied the murder bullet to a box of bullets that a witness had connected to Ware. The application also noted that state and federal courts that had reviewed his conviction pointed to that testimony when upholding his conviction. Ware v. Iowa, No. 103826 (Scott Co., Iowa, Dist. Ct.). The FOIA action The plaintiff in the FOIA suit is Washington's Forensic Justice Project (FJP). Besides collecting and disseminating information about controversial practices in forensic science, the FJP reviews civil and criminal cases to determine whether forensic science has been misused, according to its executive director, Frederick Whitehurst. Whitehurst is a chemist and a former FBI supervisory special agent who went public with criticism of the bureau's lab procedures in 1995 and later testified before the House Judiciary Committee. He was placed on administrative leave, and he later settled a whistleblower suit he had filed against the FBI. He is also a solo practitioner in Bethel, N.C. The FJP wants information from the FBI that would allow it to find the court files of the cases in which CBLA was used. The suit also seeks to obtain the results of an internal review of about 3,000 cases in which the FBI did lab work. This review, conducted from 1996 to 2004, was prompted by Whitehurst's criticisms, and it has never been made public. The FJP would notify defendants and their attorneys of the issues they may still be able to raise, and disseminate information to the general public, the pleadings say. The case docket information sought concerns the testimony of nine FBI supervisory special agents, or scientific examiners, such as Peters. He testified in Behn and about 100 other cases, 75 of them in state courts, according to documents so far produced by the FBI. Another agent, Kathleen Lundy, testified more than 80 times. In 2002, Lundy confessed to her superiors that she had perjured herself in a Kentucky murder case when she intentionally lied about the bullet-manufacturing process of Winchester ammunition. Lundy was subsequently terminated by the bureau, but not before her superiors tried to convince her that she hadn't really lied, according to interviews conducted by the U.S. Department of Justice's office of the inspector general, Case No. O & R 20022003. Lundy later pleaded guilty to a misdemeanor in Kentucky. While accusing Whitehurst and Colapinto of being "prolific filers," the FBI said in its pleadings that it has searched its indices and they do not include the names of the defendants, the names of the courts, or the docket number of the cases in which FBI employees testified, and that the data are not readily available. Not so, claims Whitehurst. "When FBI examiners testify in courts they must document that testimony in the case files, known as 'bufiles.' This practice has been in effect for 50 years," said Whitehurst. "That documentation was computerized after Oct. 1, 1986."