The Rise of DNA Felon Databanks

A Murder in Minneapolis

In the fall of 1991, Jean Broderick, a 23-year-old woman just out of Macalester College, decided to share a duplex apartment with Erica Nor-ris, one of her college friends, and two other roommates in Minneapolis. The Lowry Hill neighborhood was a bit shabby, enough to keep the rent low, but not so seedy as to make the three young women and one man worry much about safety. One Saturday night that November, Jean and Erica went out for a few drinks. When they came home they talked for a bit with Erik Helgen, the sole man in the group. That night he had seen the movie Cape Fear, the grisly tale of a paroled criminal who terrorizes a family. They talked briefly about how disturbing some of the scenes were, but the three soon went to their respective bedrooms.

Sunday morning started slowly for Erica. She got up about 11:30, tidied the apartment a bit, and then fell back asleep until the phone woke her at 1:30. The call was for Erik, but she knew he was not in because he had said the night before that he was getting up early to attend a day-long retreat for his job. Realizing how late it was and seeing Jean's door open, Erica went in to see her. Jean was facedown in bed with her hands tied behind her back. Her face was an ugly purplish color, and she was stone cold. The autopsy showed that she had been raped and strangled.

For a few horrible hours Erik was the suspect, partly because of the bizarre coincidence that he had seen Cape Fear the night before and that Erica had told the police that he had seemed powerfully affected by it. In that movie the criminal picks up a woman in a bar, brings her home, ties her hands behind her back, and brutally beats her. But Erik's alibi quickly checked out; he had been with several people during the morning hours that the coroner determined was the time of the attack. Because the other people in the apartment had not heard any unusual sounds suggesting a forced entry, detectives were convinced that the killer was an experienced criminal. When the police realized that on her last night alive Jean had walked past a halfway house for paroled sex offenders, they focused their investigation there.

In 1989 Minnesota became one of the first states to enact a law creating a DNA felon databank. In Minnesota, before he is paroled, every convicted sex offender must provide a blood sample for DNA analysis. During the mid-1990s technicians at the state Bureau of Criminal Apprehension (BCA) abandoned an older technique called RFLP analysis in favor of a newer DNA identification technology that relied on protocols perfected by FBI scientists. The standard practice was to extract the DNA and use five different DNA probes to type DNA sequences, each of which varies greatly in size among individuals. The result was a genetic fingerprint of each person. The high degree of variability in these DNA sequences is not incompatible with the rigorous evolutionary pruning of genes. The DNA used for identification lies between the stretches of DNA that code for structural genes—the ones that are blueprints for proteins. Because these regions do not code for proteins, the variations are harmless.

The detectives asked the BCA lab to compare the DNA from the semen sample that had been scraped from Jean Broderick's inner thigh with DNA patterns on file for the paroled offenders who were living at the halfway house. There was no match. The DNA evidence that they had been forced to provide at parole effectively eliminated them as suspects. There were no others. Jean had been a quiet person. She had not been dating, and none in her small circle of friends could offer a clue as to who might want to harm her.

When the police asked Joel Kohout, a seasoned detective who works at BCA, to step in, she quickly decided that the crime looked like the work of a recidivist, someone who had raped before. After agreeing that the case was going nowhere, Kohout and her lab colleagues decided to do something that had never been done before. They would compare the DNA profile of the semen sample to the DNA profiles of every convicted rapist in their new DNA felon databank. At the time, the BCA had not yet set up a computerized database of its DNA profiles, so the lab technicians had to visually compare the physical profile of the crime scene sample (essen tially, a series of dark bands on an X-ray film) to those in the database. In the first pass they were able to exclude all but six of the men who were registered in the database. Put another way, they found only six convicted sex offenders who matched the sample at the first DNA position they had picked. After running a comparison of a second DNA locus from the crime scene sample against the six remaining open files, the technicians could exclude five more.

The chances that a man randomly selected from the population of Minnesota will have a DNA profile that matches the first two DNA loci in the crime scene sample is less than 1 in 1000. But there are more than two million adult men in Minnesota, so if one tested the DNA of every one of them with two probes, about 2000 would not be excluded as suspects. This is why the FBI insisted on a multi-marker test system (in 2000 a system of 13 STRs was in use). As the number of markers that fail to exclude a suspect rises, the possibility that the match is due to chance alone falls quickly to unimaginably small numbers. This, of course, assumes that the testing laboratory has not made some gross blunder such as mislabeling a sample or testing the same sample twice, or committed fraud—the kinds of concerns that were raised by the defense at the trial of O.J. Simpson.

Jim Liberty, one of the forensic technicians, knew that he had a very strong lead, but he needed to be as certain as possible if the lab work was going to provide the sole basis for an arrest warrant. Obeying protocol, he took care not to learn anything about the sole remaining suspect. Other forensic tests had shown that the semen was from a man with a type of enzyme called PGM 2-. Liberty asked another technician to retrieve the original blood sample of the convicted felon and test it for PGM 2-, a type found in only about 1 in 200 persons. It also was PGM 2-. The chances of a randomly selected man having two DNA loci and the same PGM profile were very small, perhaps 1 in 200,000.

The Minneapolis police had made a "cold hit," a first in U.S. forensic history. By comparing the DNA profile of a crime scene sample to those in an existing database, authorities had, despite the lack of any other evidence, found their man. His name is Martin Estrada Perez, a 37-year-old career criminal. He was easy to find. He was in the Hennepin County jail where he had been since being arrested for burglary just 11 days after Jean Broderick's murder. A test of his blood showed that his DNA matched that of the DNA in the semen found on the dead woman. As the prosecutors prepared for trial, they found other evidence, including eyewitnesses who picked Perez from a lineup as someone they had seen in the neighborhood. On April 23,1993, a jury, largely on the basis of the cold hit, convicted him of first-degree murder. He was sentenced to life in prison without the possibility of parole.

The Rise of DNA Evidence

Ironically, the scientific basis for matching DNA samples from crime scenes began with the most peaceful of interests. In the early 1980s, Alec Jeffreys, a young molecular and evolutionary biologist working at the University of Leicester in England, was trying to use DNA markers to study the relative reproductive success of animals and how that related to variation in the structure of local populations. Among his first subjects were sparrows. By chance he discovered that there were many highly variable regions in the sparrow DNA that he could use for this purpose. About this time, police in England were trying to solve two brutal rape murders in two villages near Leicester. Hearing of Jeffreys' DNA identification work in birds, they wondered whether he could do it in humans. They were especially interested to find whether the DNA from the semen samples taken from the victims showed that one man had committed both crimes. Jeffreys agreed to try.

Early on, Richard Buckland, a teenage boy with a history of sexual deviance, had confessed to the crimes, but the police doubted his word. Jeffrey's studies showed conclusively that the DNA pattern of this emotionally disturbed adolescent did not match that of the semen left by the killer. Thus, the very first use of DNA testing in a criminal case exonerated a suspect who had confessed to a crime he did not commit!

The British police asked all the men who lived in or about the three local villages closest to the crime scenes to voluntarily provide a blood sample for DNA studies. More than 98% (3653 men between 13 and 34) agreed. Not surprisingly, since what killer would give blood, DNA analysis detected no sample that matched the crime scene profile. It did, however, eliminate a huge number of suspects. It also found something extremely interesting: Two of the DNA samples from the volunteers matched perfectly. Since there were no identical twins, this meant that a sample from one person had been analyzed twice. Someone had provided blood under his own name and then again on behalf of someone else. When the police approached the individual who seemed to be the source of two samples, he readily admitted that he had given blood twice, once for himself and once for a friend who had persuaded him to do so because he claimed to be terrified of needles and would be embarrassed if he fainted. The police now had a real suspect. The other man, who had prior brushes with the law for child molestation, eventually confessed.

News of the case electrified the police forces in Europe and the United States, and for good reason. Each year in the United States there are about 20,000 murders and 90,000 reported rapes (almost certainly another 90,000 go unreported). Only a fraction of these cases end in the conviction of the perpetrator; in the case of rape, much less than one-half. Yet, we know that many crimes are committed by recidivists, those who have been convicted before for the same or similar acts. For example, of convicted felons released in 1993, 62% were rearrested within three years of parole and 41% were reimprisoned.

With the exception of murder, criminals tend to commit the same crimes repeatedly during the course of their lives. This is particularly so in the case of convicted sex offenders. The high rate of recidivism provided the fundamental argument in favor of creating DNA felon databanks that swept through law enforcement circles and legislatures in the United States and Europe starting in the late 1980s. During the last decade, all 50 states have enacted laws to create DNA felon databanks. Most laws originally targeted sex offenders, criminals who comprise about 10% of all convicted felons. A few states, such as Virginia, decided early on to collect blood at parole from all persons who had been convicted of a felony. By 1999, it had become clear that most states would revise their laws to follow Virginia's lead. As a result of its decision to enact a databanking statute with broad reach, Virginia has already collected several hundred thousand samples, most of which are awaiting DNA analysis. Why? Paul Ferrara, who directs the excellent, if overburdened, state lab, only has enough resources to analyze samples from convicted murderers, rapists, and other sex offenders. Samples from less dangerous persons must wait.

In the late 1980s when courts were first considering DNA evidence, judges, who knew nothing about genetics, had three major concerns: (1) Was the underlying theoretical basis upon which the effort to match samples rested sound? (2) Was the testing technology accurate? (3) Did the particular laboratory meet the standard of care in testing a particular set of samples? Today, the first two questions have been settled in favor of DNA technology. In courts now the focus is mainly on the quality of the analysis done by a particular lab.

One of the major public policy issues in the criminal justice system today is how to gear state forensic labs up to do DNA typing on the ever-swelling number of tissue samples taken from convicted felons, but not processed. An unanalyzed sample is worthless. It provides no information to the databank. If a paroled individual rapes again, DNA analysis will not lead the police to him. In the U.S., the backlog of unprocessed and uncollected samples is already over 1,000,000 and growing rapidly. In addition, there are an estimated 180,000 "rape kits" (most of which are likely to contain semen from the rapist) stored in evidence lockers around the nation that have never been subjected to DNA analysis. Since rapists typically rape repeatedly, failure to analyze these samples is a national disgrace, and an offense against all women. In late 1999, Congress appropriated $15,000,000 to help law enforcement agencies analyze the huge backlog of unprocessed samples. This is welcome news, but we need at least another $30,000,000 to complete the job.

Despite current funding problems, DNA forensics will forever change the operation of criminal justice. England is leading the way. In August 1994, British Home Secretary, Michael Howard, announced that Britain would build the world's most comprehensive DNA forensic database. Authorized by new legislation, the Home Office has created a computerized national register of the DNA profiles of all those convicted of a felony, a master list to which every future crime scene sample will be compared. In addition, the law permits the police to construct a DNA identity profile of everyone arrested for crimes ranging from shoplifting to murder. Without obtaining consent, the police may take a blood, saliva, or hair sample for that purpose from any person who has been arrested. The DNA profiles of individuals who are exonerated are quickly expunged from the database. By extending the reach of DNA sampling to persons arrested for even relatively minor crimes, the British government has taken DNA banking to a new level and drawn fire from civil liberties groups.

In the United States there has, thus far, been relatively little debate about DNA felon databanks, probably because in most states until recently the programs have only collected samples from persons convicted of truly heinous crimes. Convicted felons have challenged the laws in at least nine states, including Virginia, arguing that they are an illegal invasion of privacy and violate the constitutional rule which forbids punishing a person pursuant to a law enacted after he was convicted of a particular crime. The courts have consistently upheld the DNA felon databanking laws as a reasonable exercise of the police power that does not violate the search and seizure provision of the Fourth Amendment. No court has yet refused to admit into evidence testimony that a DNA analysis of a sample collected from a convicted felon matched that of DNA from tissue at a crime scene.

Despite the fact that the key constitutional issues may already be settled, several contentious issues remain that are likely to be resolved only in court. For what criminal behavior is it appropriate to require admission to the DNA felon databank? Should we follow England in taking a broad approach, or does our differing legal approach to privacy require a narrower reach? After preparing a DNA analysis, should we retain the tissue sample or destroy it? The major argument favoring retention is that testing technologies evolve and that samples may have to be retested to conform to some new testing algorithm. This seems much less an issue than it was earlier on. It is highly likely that the current, federally organized system that uses 13 STRs is sufficient for at least a decade, even with the advent of more powerful tools. The major argument against retaining the sample is rooted in suspicion that the state will not protect the privacy of the samples. Theoretically, one could ask a vast number of questions about the sample, the answers to which would reveal many other facts about the convicted felon. For example, one day we may be able to ask whether he is genetically predisposed to alcohol abuse. Should the state be allowed to do that? The laws in most states currently forbid such inquiries unless the research is conducted anonymously. Another, related, issue concerns the length of time samples should be retained. Some favor a long term (e.g., 50 years), whereas others prefer a shorter period. A fair compromise might be a 10-year holding period; long enough to accommodate a new technology, but short enough to reassure those who are worried about a potential loss of privacy.

A key measure of a criminal justice system is how it treats the accused. It is reassuring that DNA felon databanking and evidence testing are a boon to the defense. When DNA testing exonerates a suspect, it does so absolutely. One early FBI estimate was that DNA testing redirects the inves tigation away from a leading suspect about 30% of the time. Even more important is the use of DNA testing to reopen old cases and free wrongfully convicted people. An FBI official who did not want to be quoted told me that he guessed that 5% of men in jail for rape were innocent of the crime for which they were convicted. In the minds of some criminal defense attorneys this is a conservative guess.

Around the country there are scores of efforts under way to reopen cases for DNA testing. Barry Sheck, one of the two DNA lawyers on the O.J. defense team, runs the "Innocence Project" at Benjamin N. Cardozo School of Law in New York. He and his colleagues are looking into hundreds of cases in which convicted rapists seek DNA testing to challenge their convictions. In 1995 he and Peter Neufeld, the other O.J. lawyer who attacked the DNA evidence, successfully convinced Westchester County to release Terry Leon Chalmers, who had served eight years for rape on a conviction based largely on the victim picking him out of a police lineup. DNA testing, which had not been done in the first trial, excluded him. In the words of Jeanine Pirro, the District Attorney, DNA is "like the finger of God pointing down saying 'You did it'" ... or, in this case, that you did not do it.

The number of incarcerated felons who have regained their freedom because a DNA analysis obtained by the Innocence Project showed that the convicted person was not the source of the semen or blood stain found on the victim is growing steadily. In 1999 Professor Sheck knew of 53 wrongfully convicted persons who had been freed, one when he was only days from execution. The number might be much higher if it were not for laws that sharply limit a duly convicted person's right to reopen the case.

The Sheppard Murder: Using DNA to Reverse Judgment

The most famous use of forensic DNA testing to help a wrongfully convicted man arose 26 years after his death. Probably ranking just behind the 1935 trial of Bruno Hauptmann for the kidnap and murder of Charles and Anne Lindbergh's son and the 1995 trial of O.J. Simpson for the murder of his former wife and her friend, Ron Goldman, the 1955 trial of Cleveland surgeon Sam Sheppard for his wife's brutal murder was one of the crime stories of the century. Sheppard claimed that on the night of July 4, 1954 he was asleep elsewhere in the house and awoke to his wife's screams.

When he rushed upstairs he was knocked out briefly by a tall "bushy haired intruder" whom, after he regained consciousness, he chased toward the waters of Lake Erie near his lakeside home. There he claimed the killer again overpowered him. But Dr. Sheppard's credibility was greatly harmed when a woman challenged his testimony that he was happily married. Susan Hayes, a hospital lab technician, testified that in the weeks before the trial she had sex with him on several occasions in his automobile.

Sheppard was convicted and served 10 years in prison for his wife's murder until he successfully overturned the conviction, largely on the basis of a review of the trial transcript, after which the appellate court concluded that the trial judge had let a carnival-like atmosphere prevail, which could have easily prejudiced the jury. A key fact that the prosecution had never adequately dealt with at the original trial was the long trail of blood that was found in the house. The prosecution said it had dripped from the weapon that Sheppard had used to bludgeon his wife and that, since Shep-pard had no wounds, it was her blood. But in 1954 there was no good means to test it.

Although Dr. Sheppard died in 1970, his son, who was only seven when his mother was killed, never stopped trying to clear his father's name. At times he has seemed close to victory. In 1996 Mohammad Tahir, an accomplished forensic scientist in Indianapolis who had volunteered to work with Dr. Sheppard's son, determined conclusively that the blood stains throughout the house were not from Mrs. Sheppard. He had been able to test strands of hair that had been taken from the bed in which she had been murdered and compare them to the blood. Since the blood was not Dr. Sheppard's and it did not match the DNA from her hair, a third person must have been present at the crime scene.

In 1959 a man named Richard Eberling, who had washed windows in the Sheppard home, came under suspicion when a diamond ring that had belonged to Mrs. Sheppard was found in his possession. He admitted stealing it, but said the theft had been from Dr. Sheppard's brother. In 1989 Eberling was convicted of murdering a 90-year-old woman after forging her will. He died in 1998 while serving a life prison term.

Although it was not brought out at trial, Mrs. Sheppard was found with her pajamas at her feet, suggesting the possibility of rape. More than 40 years after they were prepared, slides of two vaginal swabs that had been taken from Mrs. Sheppard the night of her death were examined by Dr. Tahir. He found sperm cells. Using PCR technology, he was able to get enough DNA for testing. In his report to Ohio officials, Dr. Tahir said that the testing was "not incompatible with Eberling's DNA profile." Elizabeth Balraj, the Cuyahoga County coroner, announced that she would not challenge Dr. Tahir's findings.

In a poignant coda to the Sheppard murder, the couple's son, who is absolutely convinced of his father's innocence and certain that Eberling murdered her, arranged in 1997 for his father's remains to be buried with those of his wife. At the exhumation, he obtained hair samples for DNA analysis that he claims absolutely excluded Dr. Sheppard as the source of the semen sample found on his murdered wife. There the matter might have rested had not Samuel Sheppard filed a wrongful imprisonment suit against the state of Ohio. Nothing is more embarrassing to prosecutors than an allegation that they sent the wrong person to prison. In the fall of 1999, as part of his preparation to defend the wrongful imprisonment lawsuit, Cuyahoga County Prosecutor William Mason won permission from the court to exhume Mrs. Sheppard in order to obtain tissue samples for extensive DNA analysis. Among other things, he hopes to show that Dr. Tahir's forensic work on hair purporting to be from Mrs. Sheppard is in error due to sample contamination. If this work reaches the same results as did Dr.Tahir's analysis, it will surely help the son's claim of his father's innocence.

Other Uses of DNA Databanks

In addition to championing the development of state-based DNA databanks to compare crime scene samples to archived samples, the FBI asserts that DNA databanks provide at least three other important benefits. First, in studying a crime scene, detectives can use DNA analysis to determine whether blood or semen stains derive from one or more individuals. Second, in communities that have been terrorized by a wave of unsolved crimes such as rape, by comparing samples from different crimes, forensic scientists can determine whether the violence has been committed by several different men or is the work of a serial criminal. The resolution of this issue can be extremely helpful in how other evidence is weighed. Finally, because the molecule is so stable, DNA analysis can be used to help identify badly decomposed human remains.

In the United States this kind of forensic work was first used on a large scale by the Department of Defense during Operation Desert Storm to reassemble body parts of U.S. personnel who were killed when a missile hit their barracks. It has proved extremely helpful in civilian air disasters such as the crash of TWA flight 800. In September, 1997, the DNA lab at the Armed Forces Institute of Pathology finished the task of identifying every person who died in that tragedy. The last 15 tests used DNA obtained from tiny bone fragments dredged from the ocean floor. Over the last decade, scientists have used the fact that close relatives have a high probability of sharing DNA sequences to return the bones of the "disappeared," victims of political assassinations in Chile, to their families. Since many of the victims were young parents whose small children were kidnapped and subjected to forced adoption, DNA analysis has also been used to reunite children with their true grandparents.

In 1990 the Department of Defense decided to create a set of DNA reference samples on all members of the United States military. In the solemn words of Lieutenant Colonel Victor Weedn, the pathologist and DNA expert who until 1997 directed the DNA Identification Lab at the Armed Forces Institute of Pathology in Washington, "We never want to place another set of human remains in the Tomb of the Unknown Soldier at Arlington National Cemetery." The U.S. Department of Defense now operates the world's largest DNA databank. In 1999 it contained tissue samples of more than 3,000,000 individuals. Each new recruit provides two blood samples and a saliva sample that are dried and stored separately in two locations. DNA is only actually analyzed if a problem in identifying human remains emerges. Currently, the plan is to store the samples for 50 years, but, upon discharge, a soldier may request that the sample be destroyed.

Advances in DNA forensics, such as the ability to study old evidence as has been done in Sheppard case, have caused some novel legal problems. In the vast majority of states it is quite difficult to reopen a case after final judgment. However, in cases where it is possible to perform DNA analysis on evidence that was not analyzed for trial and in which the results could exonerate a wrongfully convicted person, the individual should have the right to obtain such tests and win his freedom if the results unequivocally exclude him as the source of the DNA. We need to change the rules for post-conviction appeals to accommodate important DNA evidence, and justice demands that we do it at once.

DNA forensics also will force us to reexamine the use of statutes of limitations, the laws that forbid an individual from being tried for a crime after a number of years have elapsed. In October, 1999, with the tolling of the six-year statute of limitations on a 1993 rape only days away, Milwaukee Assistant District Attorney Norman Gahn filed an arrest warrant against one "John Doe, unknown male with matching deoxyribonucleic acid profile." Gahn found that the DNA profile in question matched that constructed from semen stains obtained from two other victims in Milwaukee who were raped about the same time. He convinced a judge that a DNA profile is better than an alias or a written physical description, both of which often provide the basis for a warrant. He has placed the DNA profile in the state DNA felon databank, and he is confident that it is only a matter of time before it turns up as a match with DNA from a contemporary crime. This will make a long-cold trail suddenly quite warm.

The advent of DNA databanks such as that of the Department of Defense, which will eventually hold tens of millions of samples, makes one wonder where we are heading. In 1998 Louisiana became the first state to enact a law permitting the taking of a tissue sample for DNA identification at arrest. Early in 1999, New York Mayor Ralph Giuliani and Police Commissioner Howard Safir publicly advocated mandatory DNA testing at arrest (a policy which if implemented would lead to 300,000 tests a year in New York City alone). In November 1999, the International Association of Police Chiefs announced that it will urge Congress to require that DNA samples be taken from every person who is arrested. Since the technology is already in place to create DNA profiles on everyone, should we do so?

For decades we have been collecting blood samples from newborns to screen them for evidence of rare, treatable disorders which, unless detected quickly and properly managed, can lead to mental retardation. It will soon be relatively inexpensive to prepare a DNA profile on each child as well. Should we? Those who favor the creation of a universal DNA identity bank argue that if we maintain a DNA profile on each person, career criminals are much more likely to be apprehended early, sharply reducing the number of crimes they commit before they are caught. Others, especially in the

United States, view such proposals as foreshadowing the rise of an Or-wellian state.

In some respects, a system to compile the DNA profile of every person would be much more fair than the current approach. In limiting DNA profiling to convicted felons we are constructing a bank whose membership will reflect current social prejudices. If, for example, blacks are more likely than whites to be convicted of a particular offense, then they will in a sense be overrepresented in the bank. The bank in turn will be of greater value in apprehending black perpetrators of future crimes than white criminals.

DNA felon databanks are here to stay. One can only guess about the future uses to which they will be put. One cause for concern is that the banks currently store whole DNA, a practice that is not necessary to accomplish the core purpose of using the banked DNA as reference samples against which to try to identify a crime sample. By retaining whole DNA on convicted felons, especially certain groups such as sex offenders or persons whose alcohol abuse led them to commit vehicular homicide, we are creating databanks that will be of immense potential interest to behavioral geneticists. If we determine that certain genetic variants are much more common among convicted felons than among the general population, it is possible that some will argue that these are markers of predisposition to commit crimes. What will we do with such information? It is likely that tests showing the presence of such genetic variants could influence decisions about parole. A parole board might well conclude that a felon with the genetic predisposition to a particular behavior is more likely again to commit crimes than a felon who does not carry the marker. Some scientists will also find it irresistibly interesting to try to identify children with such predisposing genes and track their development and behavior.

Although I support the use of DNA databanks to help apprehend criminals and exonerate innocent suspects, I oppose the use of this archived DNA for prospecting in behavioral genetics. Even in the best of circumstances, such studies are likely to yield only mildly impressive correlations. But such findings could mislead many in our society to embrace the foolish notion that crime is a function of biology, thus directing it away from the immense socioeconomic problems that breed lawlessness.

Please see print version of this book for this figure.

Prominent Glabella

Eight-year-old boy evaluated for behavioral problems and learning disabilities. The prominent glabella and long face are compatible with the presence of an extra Y chromosome (47, XYY syndrome). (Photo reprinted, with permission, from Jones 1988.) (Karyotype courtesy ofGenzyme Genetics.)

0 0

Post a comment