For an exhilarating few months in 1999, a team at Lawrence Berkeley National Laboratory's nuclear science division thought it had done something many believed impossible, synthesizing the heaviest atom yet, called element 118. They could barely believe it themselves.
A paper announcing the result was published in Physical Review Letters, the most prestigious journal in the field, and heralded in news reports throughout the world. Experimenters boldly talked of pushing further, to element 119, maybe even as far as element 126.
Then, thread by thread, the discovery unraveled. The paper was retracted, an investigation begun. By the time it was over this summer, one scientist had been fired (over his outraged objections) because of accusations of fraud, the others reprimanded (unjustly, they insist) for not being vigilant enough. And members of the lab -- once the lair of Glenn T. Seaborg, the premier nuclear scientist of his day -- were left trying to figure out how this could have happened, and how to ensure that it never would happen again.
''It's good that Seaborg died before this, because he would have been one of the co-authors,'' said Albert Ghiorso, a veteran Berkeley researcher, who holds the Guinness world record for discovering elements. ''This would have just about killed him.''
In Mendeleyev's Periodic Table, atoms are ranked by the number of positively charged protons in their core, or nucleus, where they are packed together with chargeless particles called neutrons.
In nature the most massive element is uranium with 92 protons (and, in its most common form, 146 neutrons). But scientists have learned how to use machines called cyclotrons to slam smaller nuclei into one another with such force that they fuse together. The result is an extremely heavy nucleus -- a transuranic element -- so unwieldy that it quickly disintegrates.
Berkelium, californium, lawrencium, seaborgium -- the names of some of these exotic substances testify to the expertise of the Lawrence-Berkeley scientists, sometimes called the Sheiks, for Super Heavy Element International Kollaboration or Super Heavy Element Isotope Kemists. Other atoms created at Berkeley honor scientists like Marie Curie, Enrico Fermi, Albert Einstein and Dmitri I. Mendeleyev himself.
But since the early 80's, Berkeley had been upstaged again and again by a German team at the Laboratory for Heavy Ion Research (called by its German acronym GSI) in Darmstadt. In a 15-year marathon of discovery, GSI scientists created bohrium (107 protons), hassium (108), meitnerium (109) and, in the mid 1990's, the still unnamed elements 110, 111 and 112.
''The GSI group was smarter than we were,'' Mr. Ghiorso said, ''and they had lots of backing in terms of personnel, funds and accelerator time.''
The Russians were also providing stiff competition. In 1998, a team at Dubna surprised everyone by creating element 114.
Berkeley hoped to get back in the running with a sophisticated new detection device called the Berkeley gas-filled separator, or B.G.S. Atoms would be accelerated with the lab's 88-inch-diameter cyclotron and then slammed into a target of lead. Newly fused elements would be sifted out and identified by the highly discriminating B.G.S.
Two of the lab's best scientists were involved in the effort: Dr. Kenneth E. Gregorich, the project leader, and Dr. Victor Ninov, who had come to Berkeley from GSI where he had helped discover elements 110, 111 and 112. They were looking for ways to put the new separator though its paces when they were approached by a visiting Polish theorist, Dr. Robert Smolanczuk, who had a controversial theory he was itching to test.
Nuclear scientists measure the likelihood that a reaction will occur in units called barns, which comes from the expression ''hitting the broad side of a barn.'' Colliding two nuclei in just the right way to synthesize a heavier one is so difficult that these reactions are commonly measured in picobarns, or trillionths of a barn.
''That means, in very rough terms, that you make about one atom a week,'' said Dr. Walter Loveland, a nuclear chemist from Oregon State University, who works with the Berkeley team. For creating heavier and heavier elements, the odds decrease to fractions of a single picobarn -- an atom every few weeks or less, if you are lucky. With beam time costing thousands of dollars a day, scientists were pessimistic that their machines could be pushed much further.
But according to Dr. Smolanczuk's theory, under the right conditions, one could leap across a gap of increasingly improbable atoms -- 113, 114, 115, 116, and 117 -- and generate something unthinkably large, element 118. By making some simplifying assumptions in his calculations (which raised more than a few eyebrows), he was able to predict that the chance of producing 118 was a whopping 670 picobarns, far greater than anyone had expected.
Calculations Verified 'Almost Perfectly'
When he read Dr. Smolanczuk's paper, Dr. Loveland thought the optimistic prediction was ''simply mind-boggling.'' But it wouldn't have been the first time that something outlandish turned out to be true.
''Smolanczuk suggested this strange reaction that no one thought would go,'' Mr. Ghiorso recalled. ''But because it was relatively easy, we thought, 'What the heck, we have nothing to lose.''
Over five days in early April 1999, the experimenters bombarded a lead target with a beam of krypton nuclei. The debris from the tiny collisions passed through the gas-filled separator, and various detectors recorded the energy, position and timing of each ''event.''
A result was an enormous amount of raw data that Dr. Ninov processed using software he had mastered at GSI. As the only one on the team familiar with the program, he was put in charge of the analysis.
What he was seeking was a pattern that would indicate that krypton (with 36 protons) and lead (with 82) had fused to momentarily produce a nucleus of 118, which would subsequently decay into a chain of smaller elements. For all the technology used in the experiment, the next step was decidedly low-tech. Dr. Ninov recorded his observations by hand on two sheets of yellow paper.
Several days later, he began telling colleagues that he had observed three instances of what appeared to be the decay of a 118 nucleus to form element 116 (also never before observed) then 114, 112 . . . all the way down to 106, seaborgium. It was an extraordinary claim, but there was good reason to trust Dr. Ninov's instincts.
''I had hired a world-recognized expert and we were trusting him to do the job,'' Dr. Gregorich said. A couple of weeks later, after a second accelerator run, Dr. Ninov announced that he had found another chain.
After the group had closely reviewed his calculations -- Dr. Loveland filled a binder with supporting evidence -- one chain was discarded. But that still left three good ones. Scientists at GSI also examined the results and agreed that something noteworthy might have occurred. Everyone was working from the numbers Dr. Ninov had gleaned from his own analysis. No one felt a need to go back and examine the original raw data.
The group submitted a report to Physical Review Letters, which published it on Aug. 9. The lead author was Dr. Ninov, who had 14 co-authors, including Dr. Loveland, Dr. Gregorich and Mr. Ghiorso.
''It was such a surprise that we couldn't believe it at first,'' Mr. Ghiorso told a reporter at the time. ''Robert Smolanczuk's calculations were verified almost perfectly.''
Dr. Darleane C. Hoffman, a senior Berkeley researcher and another co-author, suggested calling the new element ghiorsium.
Attempting to Confirm Experiment's Results
Before an element can enter nature's pantheon, receiving a name instead of just a number, its existence must be confirmed by other laboratories. That summer, GSI tried and failed to find a 118 decay chain. Efforts by the Riken Institute in Japan were also unsuccessful. These negative results were not necessarily fatal. Events like these are exceedingly rare, and it was possible that Dr. Ninov and his colleagues had just been luckier than the others.
Then, in the spring of 2000, the Berkeley scientists tried repeating their own experiment. This time they found not a trace of element 118. Over the next few months, an independent review committee ruled out the most obvious technical explanations -- discrepancies in beam alignment, detector inefficiencies and flaws in gathering and processing data. Perhaps, it was concluded, there had been some problem with the magnet settings. No one at this point was considering the possibility of fraud.
Meanwhile, improvements were made to the detection equipment, and in spring 2001, the Berkeley team was ready to try again. This time, the results were also disappointing until, about two-thirds of the way through the experiment, Dr. Ninov said he had found another 118 decay chain.
By now Dr. Loveland had learned to use the special software. He was stunned when the team couldn't find the pattern that Dr. Ninov had reported.
''Victor was really one of the world's experts in doing this kind of analysis and using the program,'' Dr. Loveland said. ''It was clear that something was terribly, terribly wrong.''
In the next few weeks, a second review committee scrutinized the experiments. It couldn't find any of the decay chains Dr. Ninov claimed were in the original raw data.
To get the word out as quickly as possible, Berkeley-Lawrence put out a news release withdrawing the discovery and submitted a retraction to Physical Review Letters, where the original report had appeared.
But the journal editors rejected it. Dr. Ninov had insisted to them that it was premature to repudiate the discovery before more experiments were done. He later complained that the retraction had been submitted behind his back.
By now a third committee had scrutinized the experiments, confirming that the chains didn't exist in the raw data files. The initial suspect was the analysis software, nicknamed Goosy, a somewhat temperamental computer program known on occasion to randomly corrupt data.
Over the years, users had developed tricks for dealing with Goosy's irregularities, as one might correct a wobbling image on a TV set by slapping the side of the cabinet. But a close look at the element 118 experiments found no signs that Goosy had seriously misread the data.
A statistical analysis suggested that the chance that a malfunction could randomly produce such seemingly perfect decay chains was extremely remote. It was as though Microsoft Word had crashed and, like the proverbial monkeys banging on typewriters, tossed off sentences from Shakespeare.
Another possibility was that the patterns had really been on the original data tapes but that someone, for unfathomable reasons, had edited them out. But even that would have left some kind of trace, and none was found.
Scientist's Actions Baffle Colleagues
What turned out to be the smoking gun was a computer ''log file'' -- a diary automatically generated by Goosy of everything that had occurred during the handling of the data from the 2001 run. According to this history, an analysis performed around noon on May 7 indeed showed what appeared to be an element 118 decay chain. But when the very same data were analyzed again, a few hours later, the chain was not there.
A closer look showed that it was the earlier record that had been altered; page lengths were inconsistent, and the timing of some of the events was off. In fact, investigators discovered, the events passed off as a 118 decay chain could be manufactured by cutting and pasting a few lines from elsewhere in the file and changing some of the numbers. Records from the 1999 run also indicated that at least one of the original three chains had been edited in a similar manner -- by someone using the account Vninov.
On Nov. 21, Dr. Lee Schroeder, director of Lawrence-Berkeley's nuclear science division, put Dr. Ninov on paid leave, and a week later convened the fourth and final committee, which concluded, based on what it called ''clear and convincing evidence'' that he had fabricated his celebrated findings.
To this day Dr. Ninov, who was fired this May , maintains his innocence. He acknowledges that the decay chains are not in the raw data and that files appear to have been tampered with. But he says he is as perplexed as anyone. His account on the laboratory computer system was used by everyone in his group, he says, and his password was an open secret. Any colleague, he contends, could have carried out the deception.
In recent weeks, he has been preparing a detailed rebuttal of the charges, which he says are filled with errors and contradictions. He notes that there was simply no motivation for a scientist with his impressive publication record to commit fraud -- and to do so in such a slipshod manner.
''Why create data so bad the flaws can be detected in a few minutes of examination?'' he asks. ''Why did my expert colleagues never question the obviously flawed data? Why, having apparently successfully perpetrated a scientific fraud, did I never think to delete the incriminating evidence?
''To these questions, the answer can only be because the file was not of my creation.''
Dr. Ninov's colleagues say they find the entire episode baffling. ''I used to tell people how good he was,'' Mr. Ghiorso said. '' 'Ninov is as good as a young Al Ghiorso,' I said. So when this happened I just couldn't take it.
''Why he did it, I don't know. It's a real mystery. There was nothing for him to gain, absolutely nothing, and everything to lose.''
A Retraction, And a Denouement
In the end, no one on the team emerged unscathed. As the final report put it: ''The committee finds it incredible that not a single collaborator checked the validity of Ninov's conclusions of having found three element 118 decay chains by tracing these events back to the raw data tapes.''
But members of the group say it is routine in this type of complex experiments to delegate responsibility. ''The fundamental assumption is that of trust in the honesty and competence of your colleagues, especially if they have distinguished reputations, as was the case here,'' Dr. Loveland wrote in an e-mail message to Dr. Schroeder.
On July 15, the formal retraction of element 118 appeared in Physical Review Letters. The same month, in a curious denouement that became a footnote in the Berkeley report, a paper by the GSI group was published in The European Physical Journal. For reasons unrelated to the Berkeley investigation, the scientists had gone back and repeated their previous experiments, in which elements 111 and 112 had been found. With four additional decay chains, the discoveries seemed more solid than ever.
But on the paper's last page, they noted something curious that had emerged when they routinely re-examined some of their older data. Though most of the results held up, a single element 110 decay chain from 1994 and a 112 decay chain from 1996 showed what they politely called an inconsistency.
''For reasons not yet known to us,'' the authors reported, ''part of data used for establishing these two chains were spuriously created.''
Though they didn't say so in the paper, the authors later acknowledged that the chains had been reported by Victor Ninov.