Leo Sternbach stared at the beige walls in Lab 303, trying to inv ent the scheme that would keep his bosses off his back.

It was 1954, and the Great Tranquilizer War had just broken out. Wallace Laboratories had hit the market with Miltown, and Hoffmann-LaRoche wanted a tranquilizer of its own. And it was Leo Sternbach's job to find it.

The introduction of Miltown had jolted the pharmaceutical industry, giving doctors a revolutionary tool for treating patients with anxiety. Before Miltown, doctors could prescribe barbituarates, narcotics and sedatives for such patients, but the drugs all had three major drawbacks -- they were highly addictive, they could kill in an overdose and they were impractical for daytime use because they made patients too sleepy.

Miltown had been an overnight success because, it was erroneously believed at the time, it had none of those side-effeects. And now Sternbach was under the gun to find something better -- or, at least, similar.

Take the quickest road to success, his supervisors ordered. Start with Miltown's chemical family, the meprobamates and change the molecules a little. Make them different enough to avoid violating Wallace's patent, but similar enough to produce a tranquilizer.

But Sternbach knew that simply tinkering with another man's work would be a bore. So he reached into his chemical past and came up with a compound that he know would be more fun to work with -- the benzheptoxdiazines, a substance used to make dyes.

One can wonder today, 26 years later, how the fortunes of Hoffman-LaRoche, indeed the lives of one out of every nine Americans might have been different if Sternbach had been a little less mischievous, if he had followed his marching orders.

But he did not. And his decision to go his own way on the sly would lead -- many chemicals later -- to a drug that the world would come to know as Valium.

It would, in the short space of a few years, come to represent an estimated 40 percent of Hoffmann-LaRoche's reported $1.4 billion in annual drug sales.

And it would, in that time, come to be the world's most frequently prescribed drug, a part of the lives of some 20 million American adults.

And that, say many critics, is Valium's biggest problem. For the ease with which physicians prescribe it -- in the belief that it is safe and nonaddictive -- had led inevitably its becoming the world's most abused prescription drug.

But how people might abuse -- or, for that matter, use -- the chemicals he was inventing, was the farthest thing from Leo Sternbach's mind in 1954.

"I wasn't interested at the time in helping the whole world," he says. "I was interested in working in the laboratory, to be honest."

Leo Sternbach could not possibly have viewed things differently. A chemist's chemist, his career had begun at age 12, when he had delighted in the fires he caused by mixing chemicals pirated from his father's pharmacy in Krakow, Poland. He had always been intrigued by chemistry for its own sake, rather than for the products it produced.

He had come to Nutley, N.J., by way of Hoffmann-LaRoche's Swiss headquarters, where he had been working when World War II broke out. The company had watched the gathering Nazi storm and, as Sternbach put it, sent "all the endangered species" to America -- Jews like himself and Germans who Hitler was after.

But 13 years in Nutley hadn't stifled Sternbachhs independence. He dove into the benzheptoxdiazines because he knew how to handle them in the laboratory. They were easy to work with and, in his words, "crystallized nicely." He also knew that their components added up to a moledular weight of about 200, which made it likely that they would be biologically active and might integrate some way into the human system.

Sternbach's plan was not as harebrained as it might appear. For while a research chemist sometimes knows exactly where he is headed, at other times, he simply tries to invent new compounds, not knowing whether they will turn out to be a wonder drug, a floor wax, or just another interesting, but useless, molecule.

But by 1956, Sternbach's bosses felt he had wasted two years without producing anything. He had come up with nearly 20 compounds, but not the Miltown they were looking for. And so the message came down from the top: forget tranquilizers and work on antibiotics.

Sternbach did begin to work with antibiotics, but he also continued his tranquilizer work on the side. "I was annoyed," he recalls, "because the chemistry interested me. It seemed to me there were so many possible transformations of the chemical I was working with. It seemed I had a good chance of getting something useful."

So he continued working at his blackboard, delving into his chemistry handbooks and tinkering with the compounds. He would mix different chemicals, change the temperatures at which they were mixed, alter their ability to dissolve, always changing their structure.

And, by the end of the year, Sterngbach came up with the chemical chlordiazepoxide, Roche compound No. 0609. It was Librium, Hoffmann-LaRoche's first entry into the tranquilizer market

But Sternbach waited six months before submitting the chemical worried how management would react when they found out he had ignored their orders.

"We waited until we had to clean up the lab," he recalls, and on May 7, 1957, his 49th birthday, Leo Sternbach sent No. 0609 to Lowell Randall then Roche's chief of pharmacology, for testing. The crystalline power had been found," he told his bosses, when the lab was cleaned.

In less than a week, Randall called to say the chemical was "interesting," and asked for more of it to test.

"Interesting" was something of an understatement.

To test No. 0609 for its tranquilizing effects, it was fed to mice, who were then given what is known as "the inclined screen test."

The mice were placed at the bottom of a screen, not unlike a window screen, tilted at a fairly steep angle.

Undrugged mice can climb easily to the top of the screen. But if the mice are fed a drug that is a tranquilizer, the muscles of about half of them will relax and they will slide down the screen.

What astounded the researchers about the test with No. 0609 was not that the mice slid to the bottom, but that they were alert and active by the time they got there. When mice had been fed every other drug with tranquilizing properties -- including Miltown -- they were groggy when they reached the bottom.

During other tests, the compound also passed what Roche scientists flippantely call "the cat test."

If an undrugged cat is held up by the scruff of its neck, will instinctively draw up its hind paws. But when fed a tranquilizer, the lower legs will dangle.

And when the cats were fed No. 0609, their legs tangled.

On July 26, 1957, Roche management was told that Leo Henryk Sternbach had hit the jackpot. He had found a mild tranquilizer, far less likely to cause sleep, or addiction, than anything then on the market. No. 0609 was christened Librium after the word "equilibrium."

A few months later, before he had applied for the patent on Librium, Sternbach -- much against company policy -- decided to reallly find out what he had wrought, and swallowed 50 milligrams of Librium.

According to a journal entry he made at the time, the chemist took the drug at 8:30 a.m. and was "slightly soft in the knees" from 10 to 11. His "appetite was not influenced," the journal notes, and he was "cheerful."

At 1:35 p.m., Sternbach wrote that he was "slightly sleepy," by 2:50 he was "sleepy" and by 4 p.m. he was "pretty sleepy." By 6 p.m., he was no longer feeling any effect from the pill, which was about double what would become the standard maximum single dose of Librium.

Between the time he discovered Librium at the end of 1956 and its marketing three years later, Sternbach continued to search for other members of the benzodiazepine family that might make good tranquilizers.

"You always look for something betters," he says. "You go on to cover the whole area for your patient. You don't want to have 10 compounds patented and then have somebody come along with the 11th. You explore the whole area to see how much you can change the molecule without losing the tranquilizing activity."

In that manner, by the end of 1959, Sternbach developed 7-chloro-1, 3-dihydro-1-methyl-5-phenyl-2H-1, 4-benzodiazepine-2-one.

Diazepam. It was better than Librium because it was more effective at smaller doses, and it was found, could be used for a greater variety of problems. A member of the Hoffmann-LaRoche advertising team named it Valium, after the Latin word "valere," meaning to be healthy.

From the company's standpoint, it was aptly named. In 1960, the year Librium went on the market, the firm had 3,100 employes. By 1965, two years after Valium was introduced commercially, the number had jumped to 4,300. By 1975, Valium's peak year so far, the firm had 7,500 employes.

Valium sales, spurred by one of the most intensive advertising campaigns in the industry, grew just as spectacularly. In 1964, approximately 22.5 million presciptions for Valium were filled by retail pharmacies. By 1975, that number had grown to 61.6 million.

Leo Stenbach was paid $1 for the patent to his antianxiety gold mine, a standard procedure for the transfer of a patent from a staff chemist to his employer.

But Sternbach was not left destitute. The company used to have an incentive plan, under which a chemist could earn up to $10,000 a year extra for 10 years if he made a major discovery. The plan was discontinued, Sternbach explains with a laugh, when he won the award "three or four times." He now lives in retirement on a $60,000-a-year pension, and his wife will receive half that amount if she survives him.

Sternbach says he has not been surprised by Valium's staggering success. Had someone predicted in 1960 that his new discovery would be prescribed more than 50 million times in 1978, he says he would have thought "that was marvelous and I hope the company gives me a big raise. I would not think that's crazy because if a drug's good, you'll have a tremendous amount of prescriptions."

There is little argument that, when it is truly needed -- and used appropriately -- Valium is something of a wonder drug.

It is virtually impossible to die from an overdose of Valium -- unless the drug is mixed with alcohol or other drugs. It is an effective drug against anxiety and nonaddictive if taken for short periods of time. It is a muscle relaxant. It can help with alcohol withdrawal. It helps control seizures. It is an excellent sleeping pill.

But there are perhaps hundreds of thousands of people who have serious problems with Valium, many of whom are psychologically -- or physically -- addicted to it.Sternbach says he is willing to guess that as many as 30 percent of Valium users abuse the drug -- an assertion that makes his former employers blanch. They contend that no one has good figures on the magnitude of the problem.

But if Sternbach had the chance to go back into Lab 303, in a building now in the shadow of Hoffmann-LaRoche's modern skyscraper headquarters (jockingly called the house that Valium built), he says he would not change his drug.

"I don't feel this drug has done something bad. It has done so much good that this doesn't compare with the little bad that it has caused. And, in addition, I can't get the feeling to be responsible for the drug's use. I'm responsible for the class of compounds and the [chemical] itself.

"But for its use, I mean, everything can be abused. So you cannot create things that will be abuse-proof . . . . Look how many tires [blow] away in our automobiles. Do you blame the automobile for that?"