When Science Goes Wrong (30 page)

Kim used a standard procedure to perform the analysis. First, she extracted DNA from the samples. In the case of the vaginal sample, she performed chemical procedures to isolate any sperm DNA that might be present, thus removing the victim’s own DNA. Next, she amplified the DNA in the various specimens, meaning that she increased the amount of DNA in the sample by using the polymerase chain reaction (PCR) technique. Because the PCR technique can amplify even a very few molecules of DNA, it was necessary to take care that each sample was not cross-contaminated with DNA from the other samples or from other sources within the Crime Lab.

Finally, Kim did the actual DNA typing. Using commercially-available test kits, she examined seven different sites in the genome (called ‘loci’), at each of which there are two or more different DNA sequences (or ‘alleles’) present in the population. These differences consist of varying numbers of repeats of a short DNA sequence: one allele might consist of three repeats, another of four repeats, and a third of five repeats, for example.

When six or seven loci are tested, the odds that an innocent suspect’s alleles will match those of the perpetrator merely by chance are very small – the odds are at least a few hundred thousand to one against this happening, and they may be much slimmer than that. Since there are only about 60,000 black teenagers in Houston, these kinds of odds made it unlikely, in principle, that an innocent person would be misidentified as the perpetrator of a crime such as the one Sutton was accused of.

There is one slight complication. It’s common for an individual to possess two alleles rather than just one allele at a given locus. That’s because people possess two versions of each chromosome – one inherited from their mother and one from their father. It may happen that the alleles on these two chromosomes are the same, but quite commonly they are different. If a suspect has two different alleles at a given locus, those two alleles should be present in the evidentiary sample also if the suspect is the actual perpetrator.

When Christy Kim examined the test results from the rape case, she immediately made a couple of important observations. With regard to the sperm sample from the victim’s vagina, the printout showed more than two alleles at some loci. Since no single individual possesses more than two alleles at a given locus, the sperm sample had to contain a mixture of DNA from more than one individual.

By itself, this finding was consistent with the victim’s account of having been raped by both Sutton and Adams. However, Kim also saw that, at several loci, Greg Adams possessed at least one allele that was not present in the vaginal sperm sample. Thus, none of the sperm found in the victim’s vagina came from Adams. The sperm sample on the seat of the car also failed to match Adams’s DNA.

Of course, Adams might have raped the victim without leaving sperm in her vagina or on the car seat. But in that case only one man’s DNA should have been present in the vaginal sample, rather than two, because the victim had stated that she had not had intercourse with any man, other than her two attackers, within the previous week. Given these facts, the DNA findings ruled Adams out as one of the rapists; evidently, the victim had misidentified him. Perhaps she had relied too strongly on that one item of identification, the skullcap – which is, after all, a popular form of headgear among African American teenagers.

When Christy Kim passed on the results of her DNA analysis to the prosecutor’s office, the DA quickly realised that the negative DNA findings trumped the positive identification by the victim. He dropped the charges against Adams, who was released. Nevertheless, he had been in custody for five months before this happened. Adams wasn’t the first person to suffer in this fashion: in 1996, the
Houston Chronicle
reported that another sexual assault suspect was held for nearly nine months while the Crime Lab performed a DNA analysis that ultimately proved him to be innocent. This was despite the fact that tests of this kind can be completed in two to three days.

For Josiah Sutton, the outcome of the DNA testing was quite different. With regard to the sperm sample taken from the victim’s vagina, Kim found that Sutton’s alleles were present at all seven loci that the machine tested. (There were also other alleles present that didn’t match Sutton’s alleles, but that was to be expected: they presumably belonged to the other, unidentified attacker.) And, Kim reported, Sutton’s alleles were also present in the sample of pubic hair combings, as well as in the sperm sample retrieved from the car seat where the rape had occurred.

Kim then went on to calculate how common Sutton’s DNA profile was by consulting reference tables that listed the prevalence of the various alleles at each locus for the demographic group that Sutton belonged to. She calculated that Sutton’s pattern of alleles occurred in only one in 694,000 black males. Josiah Sutton, it seemed, had left his genetic fingerprints all over the crime scene.

With positive identifications by both the victim and the Crime Lab, the DA was not about to let Sutton walk. He did enter into plea bargaining with Sutton’s attorney, but his offer – 50 years’ imprisonment – was hardly calculated to encourage a guilty plea, especially while Sutton was still loudly proclaiming his innocence. Carol Batie hired a different attorney, Charles Herbert, for her son. Encouraged by Sutton and Batie, Herbert requested that DNA samples be sent to an independent lab for testing. Kim did send the samples, but somehow they never got tested. By the time the trial got under way, in July 1999, there had been no independent review of Christy Kim’s analysis.

The OJ Simpson trial it wasn’t. The entire proceedings lasted just three days. The victim described her ordeal and pointed to Sutton as the man who had abducted her at gunpoint and raped her. Christy Kim described her analysis of the DNA. Herbert and the trial defence attorney made some efforts to challenge the believability of the witnesses – Herbert, for example, pointed out that the Houston Crime Lab was not accredited by the American Society of Crime Lab Directors, as mandated by Texas law – but it didn’t help. The prosecutor, Joseph Owmby, described Sutton as ‘evil and dangerous’, and the jury agreed. They took two hours to find him guilty.

The judge, a career prosecutor named Joan Huffman who had just been appointed to the bench, sentenced Sutton to 25 years’ imprisonment.

In prison, Sutton witnessed murder, rape, and suicide. He was injured in fights and he experienced solitary confinement. But he also converted to Islam, earned a high-school equivalency diploma and studied legal issues relevant to his case. He appealed his conviction, but lost. He wrote to the trial judge, asking for retesting of the DNA samples, but she refused.

Sutton also wrote to the Texas Innocence Network – an organisation based at the University of Houston. His application for legal assistance was rejected because the Network had a policy of not challenging DNA-based convictions. At that time, the Innocence Network, like other similar projects around the nation, viewed DNA testing as the gold standard – the ultimate arbiter of the truth that might get innocent people out of prison but would never put them there in the first place.

The first glimmer of a change in Josiah Sutton’s fortunes came in the spring of 2001, almost two years after his conviction, when the Texas state legislature enacted a revision to the criminal code that allowed convicted criminals to apply for retesting of DNA evidence under certain conditions. Early in 2002, Sutton filed a request for retesting, but his request was just one of hundreds that flooded the system. Nothing much happened, except that Sutton was moved back from prison to the county jail, where persons awaiting trial are usually held.

In the course of 2002, two reporters for KHOU-TV, the local station affiliated to the national CBS network, began to investigate the Houston Crime Lab. The reporters, Anna Werner and David Raziq, had been alerted to potential problems by defence attorneys. The reporters examined the Crime Lab’s records for numerous cases and sent seven of them for review by outside experts. One of these experts was William Thompson, a professor of criminology, law and society at the University of California, Irvine. Thompson had achieved some celebrity as a member of OJ Simpson’s defence team; he had critiqued the handling of DNA samples in that case.

When I met with Thompson in 2006, he told me that he was shocked when he studied the material he received from Werner and Raziq. ‘The first seven cases I looked at, there were egregious problems. There were outright misrepresentations of lab results: analysts would tell the jury that sample A matched sample B, and I would look at the underlying lab work and sample A did
not
match sample B. And they were failing to run proper control samples. When DNA testing is done, it’s important to run a control called a “reagent blank” [a test with no DNA] in order to make sure that you’re not contaminating these samples with foreign DNA. When you’re extracting the evidence samples along with the references samples, which Houston was doing, you could accidentally contaminate the bloodstain from the crime scene with the suspect’s DNA and thereby get him mixed in. In my own work, I always look closely at the reagent blanks to make sure they’re clean. Well, there
weren’t
any reagent blanks. In seven cases I didn’t see reagent blanks in any of them.’

Another problem had to do with the way the statistics were presented to the juries, especially in cases where the evidence sample contained a mixture of DNA from more than one person. The appropriate statistic to report is the fraction of the population whose alleles are represented in the mixture of alleles found in the evidence sample – that is to say, the fraction of the population who could have donated DNA to the sample. ‘But they were not providing that statistic,’ said Thompson. ‘They were providing the frequency in the population of people who would
exactly
match the suspect. Most other labs had figured out that in a mixture case you use what are called “mixture statistics”. So this lab was not following general practices.’

On November 11, 2002, KHOU ran the first of a series of investigative news reports about the deficiencies at the Houston Crime Lab, based in part on Thompson’s analysis. The reports drew a great deal of attention, because Harris County, where Houston is located, has long had a reputation as the death penalty capital of the world – it has executed more people since 1982 than any other city, or indeed any state except Texas itself and Virginia. Many of these cases have involved physical evidence that was analysed by the Crime Lab.

In response to the television reports and the public outcry that they provoked, the Houston Police Department commissioned an audit of the DNA lab, which revealed numerous deficiencies. In December, the lab was closed down indefinitely.

One of the people who saw the KHOU reports was Josiah Sutton’s mother, Carol Batie. She immediately contacted Werner and Raziq and told them about her son’s case. They collected the lab reports and court transcripts from the case and sent them off to Thompson for his opinion.

Thompson recounts that he and his wife, Claudia, worked on the boxful of reports together over the breakfast table one Saturday morning. Thompson studied Christy Kim’s lab report while his wife read the transcripts. One passage in the report read as follows:

A mixture of DNA types consistent with J. Sutton, the victim, and at least one other donor was detected on the vaginal swabs, unknown sample #1, debris from the pubic hair combings, and the jeans based on PM, DQA1, D1S80 typing results.

Thompson didn’t know what ‘unknown sample #1’ referred to, but by reading the transcript Claudia deduced that it was the semen stain found on the car seat where the rape had occurred. This meshed with what Christy Kim had testified in court – that the semen on the car seat could have come from Sutton. But when Thompson looked at the actual test results, he saw that this couldn’t be true: at the locus named ‘DQA1’ Sutton possessed the alleles known as ‘1.1’ and ‘2’, whereas the semen stain contained the alleles ‘2’ and ‘3’. Sutton’s allele ‘1.1’ was not present in the stain, and thus he could not have contributed to it.

Kim had actually read the tests correctly, because her lab notes recorded the mismatch between Sutton’s DNA and the semen on the car seat. But the report she wrote up for the police ignored the mismatch and wrongly fingered Sutton as a possible source of the semen. Questioned about this later, Kim said that her misstatement was caused by a ‘transcription error’.

The fact that Sutton wasn’t the source of the semen on the car seat didn’t let him off the hook, of course, since the semen could have come from the other attacker. So Thompson turned his attention to the sperm sample from the victim’s vagina. It was true, as Kim had reported, that the DNA from this sample contained Sutton’s alleles at every locus tested. At the DQA1 locus, however, the sample contained a total of four alleles: 1.1, 2, 3, and 4.1. If the other attacker’s alleles were 2 and 3, as indicated by the semen on the seat, Sutton must have contributed the remaining alleles, 1.1 and 4.1, if he was one of the rapists. But that was impossible, because Sutton didn’t possess the allele 4.1.

‘We had this sudden realisation over the breakfast table,’ said Thompson. ‘“Wait a minute,” we thought. “This makes no sense!”’