by Richard Muller
from Muller'sGroup Website
 

LUIS ALVAREZ walked into my office looking like he was ready for a fight.

"Rich, I just got a crazy paper from Raup and Sepkoski. They say that great catastrophes occur on the Earth every 26 million years, like clockwork. It's ridiculous."

I recognized the names of the two respected paleontologists. Their claim did sound absurd. It was either that, or revolutionary, and one recent revolution had been enough. Four years earlier, in 1979, Alvarez had discovered what had killed the dinosaurs. Working with his son Walter, a geologist, and Frank Asaro and Helen Michel, two nuclear chemists, he had shown that the extinction had been triggered 65 million years ago by an asteroid crashing into the Earth. Many paleontologists had initially paid no attention to this work, and one had publicly dismissed Alvarez as a "nut," regardless of his Nobel Prize in physics.

Now, it seemed, the nuts were sending their theories to Alvarez.

"I've written them a letter pointing out their mistakes," Alvarez continued. "Would you look it over before I mail it?"

It sounded like a modest request, but I knew better. Alvarez expected a lot. He wanted me to study the "crazy paper," understand it in detail, and then do the same with his letter. He wanted each of his calculations redone from scratch. It would be a time-consuming and tedious task, but I couldn't turn him down. He and I depended on each other for this kind of work. We knew we could trust each other to do a thorough job. Moreover, we had enough mutual respect so that we didn't mind looking foolish to each other, although neither of us liked looking foolish to the outside world. So I reluctantly accepted the task, as I had many times before.

The Alvarez theory had slowly been gaining acceptance in the scientific world. The astronomers had been the most receptive, perhaps because their photographs often showed large asteroids and comets floating around in space in orbits that crossed the path of the Earth. They knew that disastrous impacts must have taken place frequently in the past. Many geologists had likewise been won over. But a majority of paleontologists still seemed opposed to the theory, which was disruptive to their standard models of evolution. Alvarez took pride in the fact that some of the most respected paleontologists nevertheless liked his theory, including Stephen Jay Gould, Dale Russell, David Raup, and J.John Sepkoski.

I began my assignment by reading the paper by Raup and Sepkoski. They had collected a vast amount of data on family extinctions in the oceans, far more than had previously been assembled. That fact disturbed me; I hate to dismiss the conclusions of experts, especially conclusions based on such minute study. Their analysis showed that there were intense periods of extinctions every 26 million years.

It wasn't surprising that there should be extinctions this often, but it was surprising that they should be so regularly spaced. Alvarez's work had shown that at least two of these extinctions were caused by asteroid impacts, the one that killed the dinosaurs at the end of the Cretaceous period, 65 million years ago, and the one that killed many land mammals at the end of the Eocene, 3539 million years ago. (The age was uncertain because of the difficulty of dating old rocks.)

Astrophysics was a field I thought I knew; my work in it had earned me a professorship in physics at Berkeley and three prestigious national awards. But the paper beggared my understanding. I found it incredible that an asteroid would hit precisely every 26 million years. In the vastness of space, even the Earth is a very small target. An asteroid passing close to the sun has only slightly better than one chance in a billion of hitting our planet. The impacts that do occur should be randomly spaced, not evenly strung out in time. What could make them hit on a regular schedule? Perhaps some cosmic terrorist was taking aim with an asteroid gun. Ludicrous results require ludicrous theories.

I hurried to the end of their paper, like a reader cheating on a mystery novel, to see how Raup and Sepkoski would explain the periodicity. I was disappointed to find that they had no theory, only facts. Physicists have a wry saying: "If it happens, then it must be possible." Many discoveries had been missed because scientists ignored data that didn't fit into their established mode of thinking, their paradigm, and I didn't want to fall into that trap. Maybe it would be best to review their data, I thought, and try to judge them independently of theory. On a chart, they had plotted the varying extinction rate for the last 250 million years. The big peaks in the rate were spaced 26 million years apart.

Next I turned to Alvarez's letter. He thought there were several mistakes in the way that Raup and Sepkoski had analyzed their data. Several of the apparent peaks, he argued, should be removed from the analysis because of their low statistical certainty. Likewise, both the Cretaceous and Eocene extinctions should not be considered as part of a periodic pattern, since they were due to asteroid impacts and therefore must be random in time. This had been as obvious to Alvarez as to me. With these extinctions removed, the remaining ones were so widely separated that it looked like all evidence for periodicity had vanished.

Alvarez's approach was convincing, but was it right? It was my job to be the devil's advocate, to defend the conclusions of Raup and Sepkoski. I went back to their paper and looked at the chart again. I mustn't be too skeptical, I thought. I replotted their data, substituting the conventions of physicists for those of paleontologists. I gave each extinction an uncertainty in age as well as in intensity. The new chart looked more impressive than I had expected. It was a rough version of the one shown on page 6. I had placed the arrows at the regular 26-million-year intervals. Eight of them pointed right at extinction peaks; only two missed. The peaks certainly seemed to be evenly spaced.

Maybe they were right. I realized I had better reexamine Alvarez's case, and see if it was flawed. This job was turning out to be more fun than I had expected.

On my second reading of Alvarez's letter, I found it particularly dubious that the Cretaceous and Eocene extinctions should be excluded. How do we know that asteroids do not hit the Earth periodically? I asked. Maybe our failure to arrive at a theory just meant that we hadn't been clever enough. Not finding something is not the same as proving it is not there. I decided to reserve judgment.

A few minutes later Alvarez stopped by to see if I had finished, and I told him that I had found a mistake in his logic. It had been improper to exclude the Cretaceous and Eocene mass extinctions, I said. I presented my case like a lawyer, interested in proving my client innocent, even though I wasn't totally convinced myself.

Alvarez rejoined strongly, like a lawyer himself.

"To keep those extinctions in the analysis would be cheating," he said.

His belligerent offense threw me momentarily off balance.

"You're taking a no-think approach," he continued. "A scientist is not allowed to ignore something he knows to be true, and we know those events were due to asteroid impacts."

I knew Alvarez far too well to acquiesce in his onslaught. My approach was not no-think, I said. It was proper to ignore certain "prior knowledge" in testing a hypothesis. He had no right to assume that the Cretaceous and Eocene extinctions could not be a part of a larger periodic pattern. Maybe if we were clever enough to find the right explanation, we would see that asteroid impacts can be periodic.

Alvarez repeated his previous argument, with a little more emphasis on the phrase "no-think." His body language seemed to say,

"Why doesn't Rich understand me? How can he be so dumb?"

I repeated my old arguments. We were talking right past each other. He knew he was right. I knew I was right. We weren't getting anywhere. This was not a question of politics or religion or opinion. It was a question of data analysis, something all physicists should be able to agree on. Certainly Alvarez and I should be able to agree, after nearly two decades of working together.

I tried again.

"Suppose someday we found a way to make an asteroid hit the Earth every 26 million years. Then wouldn't you have to admit that you were wrong, and that all the data should have been used?"

"What is your model?" he demanded. I thought he was evading my question.

"It doesn't matter! It's the possibility of such a model that makes your logic wrong, not the existence of any particular model."

There was a slight quiver in Alvarez's voice. He, too, seemed to be getting angry.

"Look, Rich," he retorted, "I've been in the data-analysis business a long time, and most people consider me quite an expert. You just can't take a no-think approach and ignore something you know."

He was claiming authority! Scientists aren't allowed to do that. Hold your temper, Rich, I said to myself. Don't show him you're getting annoyed.

"The burden of proof is on you," I continued, in an artificially calm voice. "I don't have to come up with a model. Unless you can demonstrate that no such models are possible, your logic is wrong."

"How could asteroids hit the Earth periodically? What is your model?" he demanded again.

My frustration brought me close to the breaking point. Why couldn't Alvarez understand what I was saying? He was my scientific hero. How could he be so stupid?

Damn it! I thought. If I have to, I'll win this argument on his terms. I'll invent a model. Now my adrenaline was flowing. After another moment's thought, I said:

"Suppose there is a companion star that orbits the sun. Every 26 million years it comes close to the Earth and does something, I'm not sure what, but it makes asteroids hit the Earth. Maybe it brings the asteroids with it."

I was surprised by Alvarez's thoughtful silence. He seemed to be taking the idea seriously and mentally checking to see if there was anything wrong with it. His anger had disappeared.

Finally he said,

"You surprised me, Rich. I was sure you would come up with a model that brought in dust or rocks from outside the solar system, and then I was going to hit you with a fact I bet you didn't know, that the iridium layer associated with the disappearance of the dinosaurs came from within our own solar system. The rhenium-187/rhenium-185 ratio in the boundary clay is the same as in the Earth's crust. I figured that you didn't know this. But your companion star was presumably born along with the sun, and so it would have the same isotope ratios as the sun. The argument I was holding in reserve is no good. Nice going."

Alvarez paused. He had been trying to think a step ahead of me, anticipating my moves, like a chess master. He had guessed what my criticism would be and had his answer ready-but I had made a different move. He seemed pleased that his former student could surprise him.

He finally said,

"I think that your orbit would be too big. The companion would be pulled away by the gravity of other nearby stars."

I hadn't expected the argument to cool down so suddenly. We were back to discussing physics, not authority or logic. I hadn't meant my model to be taken that seriously, although I had felt that my point would be made if the model could withstand assault for at least a few minutes. He was right that I was ignorant of the rhenium discovery. Alvarez's son Walter, a geologist, had found a clay layer that had been deposited in the oceans precisely at the time the dinosaurs were destroyed.

This clay layer, the elder Alvarez hypothesized, had been created by the impact of an extraterrestrial body (such as a comet or an asteroid) on the Earth. Rhenium comes in several forms-among others, rhenium-185, which is stable, and rhenium-187, which is radioactive and disappears with a half-life of 40 billion years. In the 4.5 billion years since the formation of the solar system, approximately 8% of the rhenium-187 should have disintegrated. And, in fact, roughly that amount had. Unless the rhenium in the clay had been produced at the same time as the rhenium in the Earth (i.e., at the formation of the solar system), the ratios were very unlikely to be so nearly identical. In other words, the extraterrestrial body would appear to have been born at the same time as the sun.

Now I took the initiative.

"Let's see if you are right that the star would be pulled away from the sun. We can calculate how big the orbit would be."

I wrote Kepler's laws of gravitational motion on the blackboard. The major diameter of an elliptical orbit is the period of the orbit, in this case 26 million years, raised to the 2/3 power, and multiplied by 2. My Hewlett-Packard 11C pocket calculator quickly yielded the answer: 176,000 astronomical units, i.e., 176,000 times as far as the distance from the Earth to the sun, about 2.8 light-years. (A light-year is the distance that light travels in one year.) That put the companion star close enough to the sun so it would not get pulled away by other stars.

Alvarez nodded. The theory had survived five minutes, so far.

"It looks good to me. I won't mail my letter."

Alvarez's turnaround was as abrupt as his argument had been fierce. He had switched sides so quickly that I couldn't tell whether I had won the argument or not. It was my turn to say something nice to him, but he spoke first.

"Let's call Raup and Sepkoski and tell them that you found a model that explains their data."

So was born the Nemesis hypothesis, though I had no idea at the time where this would lead me.