Dna: The Secret of Life

by Watson, James

  The law has always had difficulty assimilating the implications, if not the very idea, of scientific evidence. Even the most intelligent lawyers, judges, and juries have customarily found it difficult to understand at first. In one famous early instance of forensic courtroom drama, blood-typing had unequivocally ruled out Charlie Chaplin as the father of a child whose mother had slapped a paternity suit on the silent-screen legend. The jury nevertheless ruled in the mother's favor.

  American courts had long applied the Frye test as their standard for admissibility of scientific evidence. Based on one of the first trials to introduce forensic proof, it tries to keep out unreliable evidence by requiring that the science on which it is based "must be sufficiently established to have gained general acceptance in the particular field in which it belongs." But being based on a poor understanding of what constitutes well-established science, the test proved an ineffective way of determining the credibility of "expert" testimony. It was not until 1993, in Daubert vs. Merrell Dow Pharmaceuticals that the Supreme Court ruled the Federal Rules of Evidence should be used: the judge in a trial should determine whether the proffered evidence is reliable (i.e., whether it can be trusted as scientifically valid).

  Nowadays, with Court TV an established part of the television landscape, and with prime-time series focusing on forensic investigations a staple of the networks, it may be hard to appreciate how difficult it was for the American legal system to swallow DNA. Though everyone had been hearing about it since our landmark discovery in 1953, it still had about it an impenetrable scientific aura. Indeed, the field of genetics seemed only more arcane every time the popular media hailed a new advance. Perhaps worst of all was the fact that DNA-supported charges were presented not as dead certainties but as probabilities. And what probabilities they were! With figures like "1 in 50 billion" bandied about to establish the guilt or innocence of the accused, little wonder some questioned the value of lawyers, judges, juries, and expensive trials when a geneticist, wrapped in the authority of science, could settle a case.

  But at all events, most trials depend on more than the comparison of two DNA samples. Meanwhile, the acceptance of the new methods progressed slowly but ineluctably. In some sense the cause of broader understanding and acceptance was aided by lawyers who made their name challenging the very cases that depended on DNA evidence. Skilled attorneys like Barry Scheck and Peter Neufeld became as knowledgeable as the experts they were cross-examining. Scheck – short, messy, and pugnacious – and Neufeld – tall, tidy, and pugnacious – gained attention searching for technical flaws in cases presented during the early days of genetic fingerprinting. The two first met in 1977 as colleagues at the office of the Bronx Legal Aid Society, a local center of legal advocacy for the indigent. After growing up in New York City, the son of a successful impresario who managed stars like Connie Francis, Scheck found his political calling when he went to college at Yale, taking part in the national student strike that followed the Kent State shootings in 1970. Ever suspicious of entrenched authority and the abuse of power, he volunteered to assist Bobby Seale's defense team during the Black Panther's trial in New Haven. Peter Neufeld grew up in suburban Long Island, where his mother still lives, not far from Cold Spring Harbor Laboratory. He was no less precocious in his leftward leanings, having been reprimanded in the eleventh grade for organizing antiwar protests.

  It was little surprise when the two young bred-in-the-bone social progressives became crusading lawyers manning the barricades of legal aid in New York City – at a tumultuous moment in the life of the city, when rising crime rates made "justice for all" seem to some an ideal endangered in the pursuit of public safety. A decade later, Scheck would be professor at Cardozo School of Law, and Neufeld would be in private practice.

  I first met Scheck and Neufeld at an historic conference on DNA fingerprinting held at Cold Spring Harbor Laboratory. The controversy was at its height in part because the forensic technology was being applied more and more broadly despite still being done with Jeffreys's as-yet-unrefined original technique, the arcane-sounding analysis of restriction fragment length polymorphisms, or RFLPs. Inevitably some results were difficult to interpret, and so DNA fingerprinting was being challenged on technical and legal grounds. The Cold Spring Harbor gathering was actually the first occasion on which molecular geneticists – including Alec Jeffreys – would confront the forensic specialists and lawyers now using DNA in the courtroom. The discussions were heated. The molecular geneticists accused the forensic scientists of sloppy laboratory techniques, of simply not doing the testing carefully enough. Indeed, in those days DNA fingerprinting in forensic laboratories was subject to little, if any, regulation or oversight. There were also challenges to the statistical assumptions, likewise unstandardized, used to calculate those imposing numbers suggesting virtual certainty. The geneticist Eric Lander spoke for more than a few concerned participants when he proclaimed bluntly: "The implementation [of DNA finger– printing] has been far too hasty."

  These practical problems were typified in a case Scheck and Neufeld were working on in New York. Joseph Castro was accused of murdering a pregnant woman and her two-year-old daughter. RFLP analysis, performed by a company called Lifecodes, had established that a bloodstain on his wristwatch was from the murdered mother. After a sustained examination of the DNA data, however, the expert witnesses of both the prosecution and defense jointly informed the judge in a pretrial hearing that, in their view, the DNA tests had not been done competently. The judge excluded the DNA evidence as inadmissible. The case never came to trial because Castro pleaded guilty to the murders in late 1989.

  Despite the exclusion of the DNA evidence, the Castro case helped establish the legal standards for genetic forensics. These were the standards that would be applied in a much more prominent case Scheck and Neufeld were to take on, one that would make DNA fingerprinting a household term in America and indeed everywhere one could find a television: the trial of O. J. Simpson in 1994. The former sports icon was facing a possible death penalty if convicted of the heinous crimes he was charged with by the Los Angeles district attorney: the gory murder of Simpson's ex-wife, Nicole Brown Simpson, and her friend, Ronald Goldman. As part of the legal "dream team" assembled by the accused, Scheck and Neufeld would make critical contributions to Simpson's defense and acquittal. Forensic detectives had collected bloodstains from the crime scene at Nicole Brown Simpson's house, from O. J. Simpson's house, from an infamous glove and sock, and from Simpson's equally infamous white Bronco. The DNA evidence – forty-five blood specimens in all – contributed, according to the prosecution's case, a "mountain of evidence" pointing to Simpson's guilt. But Simpson had in his corner the most skillful mountaineers money could buy. The challenges from the defense came thick and fast, and as the whole world watched on TV, these counterclaims would bring some of the central controversies that had been simmering for years in forensic science up to a full-blown boil.

  A decade before the Simpson trial, back in the days when prosecutors first began presenting DNA evidence, and only prosecutors commissioned the application of genetic technology, defense attorneys were quick to raise an obvious question: By what standard could one define a match between a DNA sample found at a crime scene and one derived from blood taken from the suspect? It was a particularly contentious issue when the technology still depended on RFLPs. In this method, the DNA fingerprint appears as a series of bands on an X-ray film. If bands produced by the crime scene DNA were not identical to those produced by the suspect's, just how much difference could be legitimately tolerated before one had to exclude the possibility of a match? Or how same does "the same" have to be? Technical competence came into question as well. Initially, when DNA fingerprinting was done in forensic laboratories without special expertise in handling and analyzing DNA, critical mistakes were not uncommon. Law enforcement agencies understood that if their powerful new weapon were to remain in commission, these questions would have to be answered. A new form of g
enetic marker – short tandem repeats (STRs) – replaced the RFLP method. The size of these STR genetic markers can be measured very accurately, doing away with the subjective assessment of RFLP bands on an X-ray film. The forensic science community itself dealt with the problem of variable technical competence by establishing a uniform code of procedures for doing DNA fingerprinting, as well as a system of accreditation (see Plate 51).

  Perhaps the toughest attacks, however, were launched against the numbers. While prosecutors were given to presenting DNA evidence in terms of dispassionate, seemingly incontrovertible statistics, sometimes, as defense lawyers began to argue, tendentious assumptions had been made in calculating the state's one-in-a-billion margins of certitude. If you have a DNA fingerprint from the crime scene, on what basis do you calculate the likelihood (or, more often, the unlikelihood) that it might belong to someone other than prime suspect A? Should you compare the DNA to that of a random cross-section of individuals? Or, if prime suspect A is, for instance, Caucasian, should your sample be compared only to DNA from other Caucasians (since genetic similarity tends to run higher among members of the same racial group than in a random cross-section of people)? The odds will vary depending on what one deems a reasonable assumption.

  And an effort to defend a conclusion founded on the arcane principles of population genetics can backfire, confusing jurors or putting them to sleep. The sight of someone struggling manfully to put on a glove that simply doesn't fit is worth more – much more, experience tells us – than a mountain of statistics.

  In fact, DNA fingerprinting evidence presented in the Simpson trial pointed to the accused. A blood drop collected close to Nicole Brown Simpson's body, as well as other drops found on the walkway at the crime scene, were shown with virtual certainty to be his. With an equal lack of doubt, the blood staining the glove retrieved from his home was determined to be a mixture of Simpson's and that of the two victims; the blood found on the socks and in the Bronco proved to match the blood of Simpson and that of his ex-wife.

  No, finally, in the eyes of the jury, the undoing of the forensic case against Simpson had less to do with a failure to explain the arcana of population genetics than with the old charge of police incompetence. DNA is such a stable molecule that it can be extracted from semen stains several years old or from bloodstains scraped off sidewalks or from the steering wheel of an SUV. But it is also true that DNA can degrade, especially in moist conditions. Like any type of evidence, however, DNA is only as credible as the procedures for collecting, sorting, and presenting it. Criminal trials always include the formality of establishing the "chain of evidence," verifying that what the police say was found in such-and-such a location did indeed start there before winding up in a Ziploc bag as Exhibit A. Keeping track of molecular evidence, as opposed to knives and guns, can be an especially demanding chore: scrapings from a sidewalk may be visually indistinguishable from scrapings from a gatepost, and the subsequently extracted DNA samples will doubtless look even more alike when placed in small plastic test tubes. Simpson's defense team was able to point to a number of instances when it seemed at least possible, if not probable, that samples had been confused or, even worse, contaminated.

  There was, for example, the question of the bloodstain on the back gate of Nicole Brown Simpson's house. This was somehow missed in the early survey of the crime scene and not collected until three weeks after the murders. Forensic scientist Dennis Fung presented a photograph of the stain, but Barry Scheck countered it with another photograph taken the day after the murder, in which no stain appeared. "Where is it, Mr. Fung?" Scheck asked with a rhetorical flourish worthy of Perry Mason. There was no answer. The defense was able raise sufficient doubt in the minds of the jurors about the handling and sources of the DNA samples that the DNA evidence became irrelevant.

  As we saw in the last chapter, sample contamination is one of the foremost banes of efforts to establish identity by genetic methods. Because it can yield a DNA fingerprint from even the tiniest sample, the polymerase chain reaction (PCR) is the modern forensic scientist's method of choice for amplifying particular segments of DNA. In the Simpson trial, for instance, crucial evidence included a single blood drop scraped from the sidewalk. But sufficient DNA for PCR can be extracted from cells in the saliva left on a cigarette butt. In fact, PCR can successfully amplify DNA from a single molecule, so if even the slightest trace of DNA from another source – someone handling the samples, for example – contaminates the evidence sample, the results are at best confused and at worst useless.

  In the past decade, with the broadening application and acceptance of the DNA fingerprint as proof of identity, the law enforcement community had a flash of inspiration: Doesn't it make sense to DNA fingerprint, well, everyone – at least everyone who might be a criminal? Surely, the argument goes, the FBI should have a central database of DNA records, rather as it does for conventional fingerprints. Indeed, a number of states have passed laws requiring that DNA samples be taken from anyone convicted of a violent felony, like rape or murder. For example, in 1994 North Carolina passed legislation that authorizes taking blood samples from imprisoned felons, by force if necessary. And some of those states have since extended the mandate to cover all individuals who are arrested, whether they are ultimately found guilty of a crime or not.

  The outcry from civil libertarians has been intense, and not without reason: DNA fingerprints are not like finger fingerprints. A DNA sample taken for fingerprinting purposes can, in principle, be used for a lot more than merely proving identity: it can tell you a lot about me – whether I carry mutations for disorders like cystic fibrosis, sickle-cell disease, or Tay-Sachs disease. Some time in the not so distant future, it may even tell you whether I carry the genetic variations predisposing me to schizophrenia or alcoholism – or traits even more likely to disturb the peace. Might the authorities, for instance, one day subject me to a more intensive scrutiny than would otherwise be the case simply because I have a mutation in the monoamine oxidase gene that reduces the activity of the enzyme? Some research suggests that this mutation may predispose me to antisocial behavior under certain circumstances. Could genetic profiling indeed become a new tool for preemptive action in law enforcement? Philip K. Dick's 1956 story (which inspired the 2002 movie) "The Minority Report" may not be such far-fetched science fiction as we like to imagine.

  Whatever the outcome of the ongoing debate about who should be compelled to provide DNA samples and under what safeguards these ought to be maintained, the fact is that as I write there is a huge amount of DNA fingerprinting going on. In 1990, the FBI established its DNA database, CODIS (Combined DNA Index System), and by June 2002 it contained 1,013,746 DNA fingerprints. Of these, 977,895 are from convicted offenders and 35,851 are forensic crime scene samples for unsolved cases. Since its inception, CODIS has been used to make some 4,500 identifications that would not otherwise have been made.

  One major justification for a national database is the potential for making "cold hits." Suppose investigators find some DNA – blood on a broken window, semen on underwear – at the crime scene and a fingerprint is made. Now suppose they have no leads by conventional investigative means, but when the fingerprint is entered into CODIS a match is found. That is what happened in St. Louis in 1996. The police were investigating the rapes of two young girls at opposite sides of the city, and although the two samples of semen revealed under RFLP fingerprint analysis that the same man had committed both crimes, a suspect could not be identified. Three years later, the samples were reanalyzed using STRs and the data compared with the entries in CODIS. In 2001, they found the rapist, Dominic Moore, whose DNA fingerprint was in CODIS because he had confessed to committing three other rapes in 1999.

  The interval between a crime and a cold hit can be even more dramatic, and some malefactors have been shocked to face the molecular "j'accuse" of victims long buried. In Britain, fourteen-year-old Marion Crofts was raped and murdered in 1981, long before DNA fingerprinting
was in use. Fortunately, some physical evidence was preserved, so it was possible to make a DNA fingerprint in 1999. The authorities and Crofts's bereaved family were disappointed again, this time in learning there was no match in the United Kingdom National DNA Database. In April 2001, however, when Tony Jasinskyj was arrested for assaulting his wife, a DNA sample was taken from him as a matter of routine procedure. When it was entered into the database, a match came up: Jasinskyj was found to be the unknown rapist of twenty years before.

  In the United States, crimes like rape have customarily been subject to statutes of limitations in many states. In Wisconsin, for example, a warrant for the arrest of an alleged rapist cannot be issued more than six years after the crime has taken place. Although such statutes may seem devastatingly unfair to victims – after all, does the horror of a crime simply disappear after six years? – they have by tradition served the interests of due process. Eyewitness accounts in particular are notoriously unreliable, and all memories grow hazier over time; statutes of limitations are intended to prevent miscarriages of justice. But DNA is a witness of quite a different order. Samples stored properly remain stable for many years, and the DNA fingerprints themselves lose none of their authority to incriminate.