ON THE MORNING OF AUGUST 4, 1963, the front page of the New York Times carried the headline "Dr. Ward Is Dead; Suicide Note Calls His Foes Vultures." The story revealed that the 50-year-old Ward, an osteopath and artist who had moved with ease among the upper-crust inhabitants of London's West End, had died from an overdose of sleeping pills he had consumed several days earlier. Ward's reasons for killing himself were obvious. He'd been entangled in one of the most notorious scandals ever to hit Britain, having introduced that country's secretary of state for war, John Profumo, to an attractive young woman named Christine Keeler. She subsequently had sex not only with the married Tory official but also with a Russian naval attaché, a concurrence that raised concerns about espionage. The affair almost brought down the government of Harold Macmillan, and in the affair's turbulent wake, Ward, who counted drug dealers and loose women among his associates, was placed on trial for "living off the earnings of prostitution." According to the Times, Ward left behind no fewer than 13 suicide notes. One, written in spidery handwriting to the friend in whose home he was staying, declared, "It's really more than I can stand -- the horror day after day at the court and in the streets — it's not only fear — it's a wish not to let them get me. I'd rather get myself."

In the month that followed the news of Ward's suicide, 1801 people killed themselves in the United States. Most of their deaths weren't announced on the front page of the New York Times, and if you had asked leading sociologists of the day whether Ward's well-publicized act of self-destruction had affected the subsequent American total, they would have said no. Almost 70 years earlier, the French pioneering sociologist Émile Durkheim had declared that imitation does not increase the national level of suicide; he believed that a lack of social cohesiveness instead explained why the vast majority of people kill themselves. For three-quarters of the 20th Century, most scholars accepted that.

But in the month after Ward killed himself, the number of Americans who killed themselves in fact was higher than normal -- 160 more deaths than one would have expected from looking at the numbers in the same time frame in the preceding and succeeding years. In the month after actress Marilyn Monroe's naked body was found lifeless next to an empty bottle of Nembutal sleeping pills, the number of suicides among Americans jumped 12 percent (197 more deaths than normal). In fact, spikes followed 26 out of 33 suicides reported on the front page of the New York Times between 1948 and 1967 -- a total of more than 2000 "excess" deaths. The conclusion drawn by David Phillips, the researcher who discovered this statistical pattern, was that suicide indeed can be contagious.

Phillips was teaching at the State University of New York at Stony Brook in 1974 when he published these findings in the American Sociological Review. "I was proud of the title I gave that paper," he says today. "I named this thing the Werther Effect, after Goethe's famous hero." The famous German author himself was suicidal, Phillips explains, and he wrote his novel The Sorrows of Young Werther as a way of purging himself of his self-destructive feelings. "After the book came out, all sorts of people were said to be copying the fictional hero. And I said, 'Hey, let's see if this works in real life as well as in fiction.' "

Shortly after his first Werther Effect paper was published, Phillips joined the sociology faculty at the University of California San Diego, where he remains today. Over the past 30 years, he has discovered all sorts of things relating to death. He found "that suicide stories were followed not only by spikes in suicide but also spikes in automobile accidents as well. And particularly by spikes in single-car accidents. Isn't that neat?" Phillips exclaims. It's a rhetorical question he poses often. "The more publicity given to the story, the greater the increase, and the increase came mainly where the publicity occurred. In the case of the auto data, I was able to show that after a young guy had his suicide story on the front page, the deaths of young drivers went up particularly. There was a relationship. The age of the person crashing just after the story was predictable from the age of the person dying in the story! But that was true only if you looked at the driver deaths that occurred after the first story. You could not predict the age of the passengers dying. Isn't that neat?" The ages of drivers who crashed just before the suicide story also were not predictable. And "if you looked at multiple-car crashes where presumably at least one driver didn't want to die, the age of the drivers was not neatly predictable from the age of the person who had committed suicide." In sum, all the data suggested that some of the single-car crashes were in fact suicides. "It was a beautiful, beautiful finding," Phillips enthuses.

That study was published in the journal Science. It was the first time his parents noticed that he'd published something, Phillips jokes. Both his mother and father were public-health physicians born in South Africa. Although they wanted each of their four children to become doctors too, none did, but Phillips suggests that having his work published in prestigious science and medical journals somewhat made up for that. In South Africa, "Their way of expressing their liberal politics was to go into public health and cater exclusively to nonwhite populations," the professor recalls. In the mid-'50s, when his parents had given up on the possibility of South Africa turning into a true democracy, they joined a wave of other liberal South African physicians fleeing to the United States and settled in Boston. Twelve and a half years old at the time his family immigrated, Phillips recalls that "we were never alone in that house. There were always [other] immigrating families staying with us and being settled. It was a very nice experience," and it echoed his mother's own youth, he adds. "Her parents having come from Lithuania, in all the years that she grew up in her house in South Africa, she was never alone."

Today the 61-year-old Phillips's speech reveals his birth country in subtle ways. He says "chawnce" for "chance," "cawn't" for "can't," "puh-mit" for "permit." An unprepossessing figure, his steel-gray hair has receded from a face distinguished by probing, deep-set eyes behind aviator-frame glasses. Often he closes those eyes as he talks. He says he thinks better that way. But they were open, scanning the large lecture hall within UCSD's Center Hall a bit before 8:00 a.m. on the fourth day of this year. It was the first session of Phillips's Sociology of Suicide class, a course he's taught so often he can't remember how many times it's been. Phillips was dressed in his inveterate uniform: a long-sleeve plaid shirt with a bit of white undershirt peeking out from beneath it; dark chinos; Birkenstocks over dark socks; a black pouch that he wears on his belt to store the sorts of things a woman would carry in a purse. Precisely at 8:00, he pulled out a handheld chime and struck it. "How many of you have been in my class before?" he asked. When only a few hands rose, he declared, "That's great! Then I can tell those same stories over again."

After he dispensed with the administrative preliminaries, he asked the large group assembled before him, "Why do you guys want to attend such a class?" No one raised a hand, and Phillips chided them, "If you're going to grow up and have courage to vote and so forth, you're going to need to be able to speak up, you know." One girl finally offered that she had known someone who'd killed himself, an admission that drew an understanding nod from Phillips. "But seriously, folks, if you want to worry about dying from something, you really should not worry too much about suicide," he said. "People who study suicide would like you to think you should worry about it." Such concern brings them funding, he suggested. But among college-age individuals, it happens rarely, he reassured his audience.

Another girl remarked that it was nonetheless fascinating.

"Yeah, it really is a remarkable thing, considering how many evolutionary mechanisms are tasked to keep us alive," Phillips agreed.

"I see it as part of an underground culture," a young man remarked. "There's a lot of dark material associated with the subject."

He wouldn't be able to add anything to their fund of knowledge about popular music, Phillips warned his students. Popular culture was not his forte. But he would award a prize to the student who could find the best example of a suicidal song or poem. The winner would be able to choose between a pair of ceremonial chopsticks or a T-shirt from Marshall's Industrial Hardware. "Some people want the one; some want the other, but most people want the chopsticks," Phillips said.

In the classes that followed, Phillips also gave the students a taste of some of the statistical data that's been amassed about suicide. They learned, for example, that more Americans kill themselves every year than fall victim to homicide, a fact that has remained consistent for many years. That's the case in most countries, but the gap between the murder rate and the suicide rate is smaller here than in any other industrialized nation. The American murder rate is the highest of the world's industrialized nations, while a number of other developed countries beat us out at suicide, including the United Kingdom, Germany, Japan, France, and the Scandinavian countries. The number of suicide deaths in the U.S. has been climbing since 1980, as the population has aged; older folks kill themselves at much higher rates than do younger ones. Between 1980 and 1992, 19.5 out of every 100,000 75- to 79-year-olds killed themselves, while just 8.5 out of every 100,000 15- to 19-year-olds did. If one adjusts for age, the suicide rate in America has not changed much over the course of the past 50 years, according to Phillips.

Firearms are the method of choice, accounting for 60 percent of all suicides in 1992, followed by poisoning (18 percent), strangulation (15 percent), and cutting (1 percent). No matter what the means, by far the most likely suicide candidate is male, with white males doing themselves in at far higher rates than their brethren from most other races (Native American males being the grim exception). White females kill themselves at a higher rate than blacks, Hispanics, Native Americans, or Asians. Among the factors that appear to be protective against suicide are education (at least until later in life), marital status (the rate of suicide among married people being less than half that of those who are single, widowed, or divorced), and geography (with most of New England and the Midwest having lower suicide rates than the rest of the country).

The reason we know these things is because when people die in the United States, we record a lot of information about them on their death certificates -- including their sex, race, age, marital status, level of education, where they were living at the time they died, and what they died from. "Because a death certificate is a matter of legal importance, some pieces of it are filled out very accurately," Phillips says. "I think this society keeps good records on things that are important to it -- or scary."

What many people don't realize, he adds, is how detailed the cause-of-death information is; a coding system allows state authorities to choose among some 10,000 possibilities. Thus it is not merely recorded that a given person poisoned himself. There are separate categories for killing oneself with alcohol; with "narcotics and psychodysleptics"; with a pesticide; with "non-opioid analgesics, antipyretics, and antirheumatics"; with "antiepileptic, sedative-hypnotic, antiparkinsonism, and psychotropic drugs"; and so on. You don't just die from a gunshot. You die from a handgun shot, a rifle shot, a shotgun shot, or some other kind of firearm -- and the death certificate distinguishes which it was.

Every year millions of dollars are spent and millions of person-hours invested in collecting these death certificates and putting them in computer-readable form and sending them to the National Center for Health Statistics. "There are two million deaths a year," Phillips points out, "and the feds provide this stuff from 1968 till now. So that's more than 60 million records. It's a huge data set."

Yet almost no one studies the information that's collected, he marvels. "Some people seem to think there's no point in looking at data that's precollected," he says. Phillips doesn't understand that. He says, "With this kind of precollected data, the quality is very much more tightly controlled than when you collect it yourself. When you collect it yourself, you don't have an army to do studies of the quality of the data and to improve it. But when the feds and the states do this, they've got people year after year after year [examining the quality]. That's all they do."

If you want to understand something like suicide, it might be tempting to employ a more direct approach, say interviewing people who tried but failed to kill themselves or talking to their surviving friends and relatives. But the more indirect approach of studying the precollected data can yield enormous insights, Phillips believes. "I think of what I do as kind of like what an astronomer does -- or what astronomers did before they sent up these probes," the professor says. "For 2000 years, they said, 'Okay, we are restricted to the data that are available. We cannot do experiments.' And look at how much was found out!"

Phillips says his personality was another factor that impelled him to his first big finding about imitative suicide. "I'll teach you a word you probably have not encountered," he offers when explaining this. "It's called davka. It's a Hebrew word; also a Yiddish word. And it's used like this. 'He told me to go uptown, so, davka, I went downtown.' " Phillips says Samuel Johnson put his finger on this quality when he wrote that some birds can fly only against the wind. "That's me and Samuel Johnson both," Phillips says. "The great god Durkheim had said there was no such thing as imitative suicides, so I said, 'Let's see if there's such a thing as imitative suicide.' "

After he discovered the Werther Effect, Phillips began wondering if more than just news reports of suicides might prompt certain individuals to kill themselves. He wondered if fictional suicides had a similar effect, as had been reported following the publication of Goethe's novel. Very few studies of this question had ever been done, he discovered, so he came up with a plan. "I found a soap-opera magazine which listed soap-opera plots." Studying that, he came up with a timeline for when various fictional TV characters had killed themselves on the airwaves. "You know, there are people who identify like crazy with these soap-opera characters," he points out. "In some cases, people track the fictional life more closely than they track anybody else's life."

Phillips says when he looked to see whether suicide rates in the United States went up after the soap-opera suicides, he did find a correlation. "But it was just a couple of years' worth of suicide stories, so it was not a lot of data." He says a much better study was done by some German colleagues who looked at what happened in Germany after the airing of a four-part miniseries. Each part of the series opened with the same ten-minute segment in which a suicidal young man lay down in front of a train and got run over by it. "My German colleagues said to the television network, look, are you aware that there are studies by Phillips showing that this is a potentially dangerous thing? And the [German] network said, too bad. Freedom of the press." Phillips isn't criticizing freedom of the press. He points out that "in South Africa, we didn't have freedom of the press, so I value it highly."

But the fact is that the German researchers were able to show that in the months after the fictional railroad suicides, actual railroad suicides climbed significantly. "And that was particularly true for young males," Phillips says. "Then a year later, this television network decided to show this miniseries again, and my colleagues again pointed out the findings." When they were ignored, they once again studied the consequences and found another railroad-suicide spike.

Phillips says an Austrian study has shown what happens when the press does pay attention to the impact of publicity about suicides. "People were jumping from the subway platform in front of subway cars. And this was being written up on the front pages of the newspapers. These researchers went and said, 'Look, are you aware that studies by Phillips and other people show that when you newspapers publicize this stuff big-time, there's this spike in subsequent suicides?' And the publishers in Vienna, who were the main publishers for all of Austria, said, 'No, we didn't know that.' " Unlike the German TV programmers, they agreed that a change in policies might be worthwhile. "They said, 'Okay, we will no longer carry such stories on the front page. We'll bury them on the back page,' " Phillips says. They also stopped running photos with the stories and tried to avoid using the word "suicide" in the headlines. After the two biggest newspapers in Austria changed their pattern of reporting, "What the researchers showed was that the number of subway suicides dropped precipitously." It was a different way of studying imitative suicide -- but it confirmed his findings, Phillips believes.

In the course of his career, Phillips has discovered other patterns relating to suicide. When he looked at Californians who killed themselves between 1969 and 1985, he found that many more married men did themselves in at ages ending in "0" and "5" than one would have expected. "For many people, 'boundary ages' of this sort may serve as a signal that one promising period of life is over and another less promising one is about to begin," Phillips has written about this phenomenon. Particularly lethal were the ages of 40 (at which 14.67 percent more married men killed themselves than one would have predicted) and 65 (which had a 14.18 percent higher rate). Those two "symbolic ages" are "traditional boundaries between youth and middle age, and middle age and old age," Phillips writes. "At these boundary ages, a man may reflect on whether or not he has achieved his goal, has properly provided for his family, and has good prospects for the future." Some of those taking stock "are likely to consider that they have failed themselves and their families."

Phillips also has found that acts of self-destruction do not increase during major holidays, as some researchers had suggested. "They were mainly getting their information from very small, geographically limited samples," he says. When Phillips looked at data collected from all over the United States, he found a large drop in suicides just before a number of holidays, followed by a small peak in suicides immediately afterward. "The net effect of the holidays on suicide is actually a drop," he concluded.

It isn't just suicides that appear to decline before holidays. In a series of studies that have gained international attention, Phillips has found evidence that some people dying from natural causes seem to be able to hold the Grim Reaper at bay until after the passage of various important occasions. When he looked, for example, at the death rate among Chinese people living in California between 1960 and 1984, he discovered that old Chinese women die at a much lower rate than expected just before the Harvest Moon festival, and they die at a greater-than-expected rate in the week after the holiday. He could not find this "dip/peak effect" among old Chinese men or younger Chinese women. But the festival is of greatest importance to the Chinese matriarch, who "takes charge of preparing an elaborate meal and, by custom, directs the work of the daughters and daughters-in-law." Phillips concluded that the dip/peak pattern demonstrated that people could briefly postpone their deaths until after a significant occasion. (He has also documented the same phenomenon among men with Jewish surnames just before Passover and among Californians right before their birthdays.)

Christmas and New Year's appear to have a different effect on mortality. Phillips says he found this out after he began wondering whether the two big end-of-the-year holidays might have something to do with the fact that mortality from heart disease rises during the winter. "Nobody had really reflected on this topic before," he says.

To test the hypothesis that people don't have more than the normal number of heart attacks during the Christmas/New Year's holidays, he calculated how many deaths from heart disease would be expected to occur on every day between July 1, 1973, and June 30, 2001 (taking into account the effect of seasons and trends on mortality). Then he compared those projections to 53 million deaths from heart disease that in fact were recorded between July 1, 1973, and June 30, 2001. The result was that a sharp spike in deaths from heart attacks could be seen during almost all of the two-week periods from Christmas through January 7 that Phillips included in his study -- many times more than should have been seen if the holiday period had exerted no deadly impact.

Why does that spike occur? Phillips says, "What I primarily do is reveal the existence of phenomena that were not previously known, rather than providing in great detail the explanation for those phenomena." He says other researchers can go through the doors his work has opened to investigate the causes in depth.

At the same time, it's clear he enjoys thinking up potential explanations for the patterns he finds and looking for ways to prove or disprove those explanations. He makes a point of doing this in all his scientific articles. In his first big imitative suicide paper, for example, Phillips considered the possibility that the reason the American suicide rate jumped after stories about suicides ran on the front page of the New York Times was not because those who killed themselves were being copycats but rather because the stories about the suicides made them sad. "Most persons whose suicides were reported on the front page of the Times were not widely known until they killed themselves; in addition, many were in trouble with the law and thus not likely to be admired by the general public. Consequently, it is difficult to believe that these persons could elicit sufficient grief to increase the national level of suicides," he wrote. Still, in order to test that possibility Phillips examined the suicide statistics for the months following the deaths of eight American presidents, between the years 1900 and 1968. "In general, presidents are among the most famous and admired persons of their day. If the Werther Effect is caused by grief at a publicized death, then U.S. suicides should increase significantly after presidential deaths," he reasoned. In fact, he found no such statistically significant tendency.

More recently, Phillips considered ten possible explanations for the spike in heart-attack deaths during the Christmas/New Year's holidays. He found that most of those explanations didn't appear to be supported by the data. For example, to test the notion that changes in diet and alcohol consumption associated with the holidays were causing the spike, he looked at the heart-attack death rate among hospital inpatients, whose diet and alcohol consumption are controlled. He found a death spike among their numbers, and he also discovered that "people dying from cardiac diseases with substance abuse listed as a secondary cause of death produce a smaller holiday peak than do people dying from cardiac diseases alone" -- more evidence that overindulgence is not the culprit.

To see whether the cause might be emotional stresses associated with the holidays, he focused on Alzheimer's disease patients. Such individuals tend to be less aware of holidays than people without Alzheimer's disease, he pointed out. If the emotional stresses of Christmas and New Year's cause heart attacks, the mortality of the Alzheimer's sufferers should be less affected by the holiday period than other groups are. But instead of dying at a lower rate, the Alzheimer's patients' rates were higher.

Phillips thinks the explanation that holds the most water is that the health-care system tends to function abnormally at this time of the year. Individuals may put off seeking medical attention because they're traveling or they don't want to disrupt the festivities. Indeed, previous studies had shown that admissions to urgent-care facilities do drop on holidays and spike right afterwards. Phillips reasoned that if this delay-in-seeking-care hypothesis were correct, other natural deaths should also be higher than expected during the two-week period, and he did find such a peak among inpatients, outpatients, those who were dead on arrival, and those who died in the emergency departments.

Moreover, patient behavior is probably not the whole explanation, he adds. Other changes in the medical-care system also take place. Doctors and nurses and pharmacists take holidays then, as do employees in the insurance industry. Those sorts of disruptions may explain why Alzheimer's disease patients and nursing-home residents die at higher rates during the Christmas/New Year's period, he suspects.

All in all, Christmas is a much more difficult holiday to study than the holidays he'd studied in his earlier death-postponement research, Phillips says. Both Passover and the Chinese Harvest Moon festival met two important criteria: they move around the calendar "so you can see whether the drop in deaths moves around with the occasion. And second, the occasion is of great interest to one group of people and not to most others -- so you can see if the people who care about it produce [a given effect] and the people who don't care about it don't produce it."

Christmas meets neither of those criteria. It's always on the same day, and furthermore, it affects everyone, Christian and non-Christian. "Many things change at Christmas," Phillips points out. "Since any change in mortality around Christmas may be influenced by changes in staff as well as changes in patient motivation, you really don't know which is which. It's too murky.... It's like looking at a pond into which a large number of raindrops are falling nearly simultaneously. You can't tell which ripple comes from which raindrop."

Phillips ran into a different sort of problem when he began to grapple with a question about murders. He says after his first big copycat suicide article was published, reporters who interviewed him asked whether there was such a thing as a copycat murder. He says he began to ponder that, but he quickly realized he couldn't use the same methodology he had used for the suicide story -- that is, see whether murder rates went up in the wake of heavily publicized murders. That's because although there are more suicides than murders in the United States, far more stories about homicides, both fictional and real, fill the airwaves and the pages of American newspapers. "Hardly a day goes by without a homicide story," Phillips notes. "Whereas in the case of suicide stories, there might be two big ones a year." (Maybe people feel more threatened by murders than by suicides, he speculates.)

Phillips says he doesn't just sit down at his desk and think up a solution to this kind of research challenge. "It comes to me from my unconscious," he says, "and then later my conscious mind writes it down. A lot of great discoveries come this way."

He says the inspiration for how to tackle the copycat-murder question came to him when he was teaching a course on the mass media and violence. Part of the research on that subject includes a study that showed that nursery-school children were more likely to hit a doll if they had watched a cartoon character acting in an aggressive manner. From that and other research, some reviewers had concluded people were most likely to copy aggressive behavior in the media if it was presented as real, exciting, rewarded, and safe from criticism. "So I said to myself, 'Well, gee, in our culture, what violence is presented as real and exciting and rewarded and not criticized?' " Phillips recalls. The answer he came up with was heavyweight boxing. "Within the ring, you can knock somebody unconscious. If you did that outside the ring, you'd be put in jail, but inside the ring, you get millions of dollars for doing it."

Phillips thus looked at the aftermath of heavyweight championship boxing matches. "I had many years of data for this, and I found several things. One is that homicides do go up in the week after those stories." Secondly, the more publicized the fight, the bigger the increase in the subsequent murder rate. He says what was still more telling was that the race of the person beaten in the fight tended to be the same as the race of the people murdered afterward. "So if a young black male was beaten in the fight, the murders of young black males went up significantly but not the murders of young white males -- and vice versa."

Another question relating to murder that Phillips has gotten involved with is whether capital punishment tends to deter would-be murderers. For inspiration on how to find evidence that would help answer this question, the sociologist looked to one of his greatest heroes, the great British epidemiologist John Snow, discoverer of the mechanism by which cholera is transmitted. ("Almost certainly you're alive because he saved ancestors of yours who lived to breed and produce you," Phillips lectures his students about Snow. "So you should be properly grateful to him!") Phillips knew that Snow had relied in his research upon something called the registrar general's weekly returns -- a weekly listing of the number of deaths, along with their causes, in London. "I knew that the Brits were still publishing this information today...just for London. The Brits are like that," Phillips says. "And I also had read lots of Sherlock Holmes and other classic murder stories. I knew that in England, the time between the crime and the apprehension and execution of the criminal could be a very, very brief period, as opposed to what happens now in the United States."

Reasoning that "if anything showed the deterrent effect of capital punishment, it should be the British crime statistics," Phillips looked at British crime committed in the period of (roughly) 1875 to 1925. "The first half of that period in England, executions were public. So I said, 'Okay, let's see if homicides go down just after this heavily publicized execution.' And indeed they did go down," he says. "That was the first time anyone had ever found that. They went down by about 35 percent. It was amazing."

But when Phillips looked at the crime statistics a bit further down the road -- say two to three weeks after the sensational executions -- he also found that "then the murder rate went up above normal by another 35 percent." The conclusion he reached "was that the lesson of the scaffold is real -- but just temporary. I mean, look, in those days, a lot of the murderers were guys killing their wives or girlfriends." Phillips conjures up a mental picture of such a fellow. "He's been irritated by this woman for 30 years. He's about to do her in, and then this dreadful example gives him pause. But three weeks later, he forgets it. Children are like that. We're all like that. It's easy to be scared off in the short run, but not in the long run."

Phillips says after that paper was published, "I was popular with neither the liberals nor the conservatives, for obvious reasons." He likewise became the target of attack by the gaming industry in 1997 after he published a study linking gambling and suicide. For many years, Nevada has had the highest suicide rate in the nation, with Las Vegas ranking as the nation's suicide capital. But Phillips did more than just note that. He demonstrated that the suicide rates among both visitors to and residents of Las Vegas, Reno, and Atlantic City were much higher than normal. He also showed that "in Atlantic City, abnormally high suicide levels for visitors and residents appeared only after gambling casinos were opened. The findings do not seem to result merely because gaming settings attract suicidal individuals."

Although Phillips defended this work against its subsequent detractors, he sounds sympathetic to their complaint. "Nobody likes to have it said that what he does for a living is dangerous for other people, right? If somebody said to me, 'David, it turns out that your teaching is very bad for people,' I would resist that vigorously. And irrationally."

He also concedes that his findings were "only suggestive." He says, "All these things that are being studied here are triggers. You're not studying the whole gun. Something like suicide is very complicated." And the relationship between gambling and suicide is particularly hard to decipher. "On the death certificate, it doesn't say that the guy's a gambler or not a gambler. Furthermore, only a fraction of people who gamble are problem gamblers."

The trickiness of studying gambling's impact on suicide rates didn't discourage Phillips from wondering if suicide rates for the San Diego area go up during the Del Mar racing season. He says he created a calendar of a dozen or so years' worth of California suicides, organized by day, and he was paging through that calendar on his computer screen, looking at the data in the company of one of his undergraduate students named Natalie. "Natalie was much more davka than I," Phillips says. "Much more than anyone I ever met. So when I said, 'Let's see if the number of suicides goes up when the Del Mar racing season opens,' Natalie, being Natalie, wasn't paying any attention to me at all." Instead she was noticing that at the beginning of each month, the number of suicides almost always was a three-digit number, rather than a two-digit one. Startled by this observation, Phillips suggested that they find out whether the pattern was consistent; they wound up analyzing data from all the death certificates in the United States between 1973 and 1988 -- 31,976,612 deaths in all. What they learned was that the equivalent of roughly 4320 more deaths occurred in the first week of each month than in the last week of the preceding month in an average year. "This monthly pattern is as regular as a heartbeat," Phillips says.

He published the results in 1999 in the New England Journal of Medicine. That article points out that the largest increases in the numbers of deaths at the beginning of the month were for deaths involving substance abuse (according to the telling death-certificate codes). Maybe people were receiving welfare or other government payments on the first of each month, using some of the money to buy illegal drugs and alcohol, and suffering the consequences, the paper suggested.

It was a reasonable assumption, Phillips says today. But it overlooked another possibility that became clear to him a few years later, as he began to be aware of (and publish articles about) the fact that the number of fatal medication errors in the United States is skyrocketing. (It nearly quadrupled between 1979 and 1998, an increase far greater than the percentage increase in the number of prescriptions filled.) When Phillips looked at beginning-of-the-month medication-error deaths that occurred between 1979 and 2000, he found that the spike in that number (25 percent above normal) far exceeded any of the other beginning-of-the-month death spikes. He now speculates that some people wait until the beginning of the month to purchase their medicines, thus increasing pharmacy workloads and leading the harried, overworked pharmacists to make mistakes that in many cases are killing people. "That doesn't undo the idea that in addition there's an increase in deaths from street drugs, which are occurring for related but different reasons. Those people are also getting their money at the beginning of the month, and they're also buying extra drugs. In both cases, the groups buying the drugs were more able to do so because they got extra money at the beginning of the month."

For his first paper on the Beginning-of-the-Month Effect (as he has called it), Phillips included his student Natalie as a coauthor. Through such collaboration with him, more UCSD undergraduates have had their names appear on articles in prestigious medical and scientific journals than they have with any other UCSD faculty member, Phillips boasts. To every class he teaches, he makes a pitch for his students to work with him on independent study projects. This benefits both him and them, he tells the students. They get credit for the articles that result from this work, while he learns a lot from them. "This is a typical example," he says. "My father died in December, and in his last words to me, he used a Yiddish word which I need to explain. It's truly a wonderful word. It's naches -- which explicitly means the pride and pleasure one gets from one's children. Isn't that a wonderful word? So my father's last words to me were, 'You've brought me a lot of naches.'" Phillips told one of his classes about this, and they peppered him with questions about the word. "So I spelled it, and I asked if any of the languages represented in the class -- there were probably 15 -- had a word for the pride and pleasure one feels in one's children. They said no." He thought that was fascinating.

Sometimes his students lead him to ideas he would not have come to on his own. Natalie did that, as did a group of Chinese students he was driving with down Clairemont Mesa Boulevard one day. When they passed the 99 Ranch Market, Phillips wondered aloud why the Asian food store was called that, and the students told him that the number nine had a very positive connotation in Chinese. Phillips asked if any numbers had sinister connotations, "and these guys said the number four had negative implications in Cantonese, Mandarin, and Japanese."

He was thrilled to hear that. For years he had been wondering whether heart-attack deaths go up on scary occasions, but he had failed to think of a good way to study that question. "You don't want the occasions to be genuinely scary," he says. "Because if they're genuinely scary all sorts of other things change too. For example, people had previously shown that heart-attack deaths went up when there were earthquakes. Those are scary occasions. But they're also occasions when the flow of medical care is disrupted and when the roads are disrupted. So the fact that heart-attack deaths go up during earthquakes need not at all imply that this is because of extra fear."

Armed with the tip from his Chinese students, Phillips checked the mortality rate for individuals of Chinese and Japanese origin. He found that it was higher than expected on the fourth day of each month, whereas Anglo deaths showed no such spike. He dubbed this the Hound of the Baskervilles Effect (after the Sherlock Holmes story in which extreme psychological stress causes the character Charles Baskerville to have a fatal heart attack) and published it in the British Medical Journal in 2001.

For as much as he delights in such discoveries, Phillips says teaching brings him an equal amount of pleasure. He thinks his students are as bright and challenging as they ever were, though the attention span of the current crop may be shorter than that of earlier generations. "I see some people twitching, and I see some people yawning," he announced on the second day of his Sociology of Suicide class. "How many of you in the course of your childhood watched only public television?" he asked. When four or five hands went up, he remarked that those individuals probably found it easier to sit still for a half hour at a time. On network television, he informed the assembly, the commercials come every 12 minutes.

So he would offer them his own version of a commercial break, he announced. "Every class I will read to you a great poem. I'm going to start with an easy poem because many of you are not very familiar with poetry. I was not very familiar with poetry until my mother died." She was a professor of medicine at Duke University, he told the class, and before that, in South Africa, she had studied the feeding patterns of infants of different races. "One day she told me that when rhinos wean their young, they immediately eat thorn bushes. Can you imagine that? They go from mother's milk to thorn bushes!"

The poem he would read them that morning was a "mother's milk poem," Phillips told the group, adding, "You thought I was just talking to you about my mother!" He then read them Tennyson's "Charge of the Light Brigade" and used that to initiate a brief discussion of altruistic suicide and the way behaviors can range from nonsuicidal to fully suicidal. Later poems would get more and more "thorn-bushy," he promised.

Phillips says he's added more and more poetry to his Sociology of Suicide class over the years. He says the students don't know what to make of it at first, but they often come up to him later and comment on how much they liked the poetic diversions. He tosses them other offbeat intellectual tidbits too. He might be instructing a class on how to pass around the attendance sheet (going along one entire row in one direction, then going the other direction in the next row) when he'll announce that this pattern is known as a boustrophedon, from the Greek for "as the ox plows." This will lead him into a mini-lecture on alphabets and directionality. "Isn't that interesting!" he'll exclaim. "You'll probably remember nothing else from this class, aside from that!"

He'll mention the Geisel Library and toss in the aside that one meaning of the word geisel in German is "an instrument of torture." He'll share information about his family, the foundation of his life. (Anyone who looks up Phillips's curriculum vitae on UCSD's website sees that the first thing he lists under his name is "Marital Status: Married [two children]," set in larger type than he uses for listing his degrees from Harvard and Princeton.) He doesn't hide the fact that he idolizes his wife of 40 years, Juliet. Thinking of her reminds him of the line in John Masefield's famous poem, "Sea-Fever." All I ask is a tall ship and a star to steer her by. His wife "can steer by a star," according to Phillips. "I'd just say, 'Oh! Here comes the wind. Boom! I think I'll go this way.' But she's got this long-distance view of where to go, and it's made a huge difference to our family."

Phillips allows that his own view of his research has broadened over the years. He thinks in the earlier part of his academic career, "I was just interested in being clever." His cleverness and the fact that "everyone is interested in death" have won him a lot of attention -- interviews by newspapers and television crews from countries all over the world -- and such press coverage is more important in academia than one might imagine. He once published an article in the New England Journal of Medicine showing that if an article from the New England Journal of Medicine gets discussed in the New York Times, it subsequently gets cited by other scientists more often than articles from the same issue of the New England Journal of Medicine that were not covered by the New York Times.

"Now, you might think the New York Times was merely flagging scientific articles that were of very high quality anyway. In other words, they weren't actually creating any buzz; they were just reflecting buzz," he says. "But we tested that in the following beautiful way, I thought." Phillips found a period in which the New York Times was on strike but continued to produce an edition of record -- papers that contained the real news of the world but were never distributed to readers. During that interval, "They were continuing to decide which New England Journal of Medicine articles they would cover and which not. So if the increased scientific visibility was occurring merely because the New York Times was good at flagging high-quality articles, then the articles they flagged during the strike should have also gotten abnormally large amounts of subsequent scientific attention," Phillips says. "On the other hand, if the New York Times was actually creating some of this visibility, then articles that they flagged during the strike period should not have fared particularly well. And that's what turned out to be. During the four-month strike period, the articles flagged by the Times did no better than the articles not flagged by the Times. Isn't that a pretty design? Yeah, I was pleased with that."

Phillips says in recent years, he's become "much more aware of the value of doing stuff which might make a difference in policy." Like his recent article about the increase in deaths over Christmas and New Year's. He'd like to think it might make people a little more careful when they travel during the holidays. His article about the beginning-of-the-month medication errors caught the attention of the Congressional Quarterly, a fact that gives him hope it might somehow influence public policy.

As for his findings about imitative suicide, he allows that maybe they've made some difference, as the findings of that one Austrian study suggested. The folks at CBS News once called him just before airtime, seeking advice about how to play the news that the Pennsylvania state treasurer had killed himself with a gun on-camera. (The professor suggested they refrain from airing images of the actual shooting, which the network did.)

Phillips has written suicide-coverage guidelines for the World Health Organization. He's talked on one occasion to the Associated Press Managing Editors association, and he says he's heard that some news media have changed their policies as a result. So maybe his work has made some difference, he allows. "But it's of course not measurable," he hastens to add. "I once asked my mother how many lives she had saved. And she said she couldn't definitively say that she'd saved any. So if she couldn't say that, how could I?"