It all began back in 1981. On the third day of August a Delta rocket lifted off from Vandenberg Air Force Base carrying a pair of NASA satellites, both known as Dynamics Explorer, into elliptical orbits. One of these satellites circles the poles of Earth at an altitude ranging from 350 miles to 14,500 miles. I was responsible for three of the instruments aboard the satellite, one of which is an ultraviolet camera, built and operated by my colleague John Craven.

This satellite was designed to examine Earth for certain light emissions that are invisible to the naked eye. I and other scientists hoped that these emissions would provide further insight into the nature of the auroral lights that occur in Earth's polar atmosphere and detect any effects associated with them. I was particularly interested in getting the first global pictures of Earth's aurora and I was not disappointed.

The pictures sent back from the ultraviolet camera on the satellite were spectacular. The remarkable auroral crowns encircle the poles of Earth, while the planet's dayside looks like a bright ball illuminated by a flashlight. This bright feature is known as the dayglow. Dayglow is produced by the interaction of sunlight with the atomic oxygen present in Earth's upper atmosphere. The ultraviolet light emitted by this dayglow is not visible to the naked eye but is within the range of the satellite's specially-designed camera. The emissions it captures are transformed into a normal photograph.

But the images of Earth we obtained beginning in late 1981 contained an unexpected feature. The blanket of dayglow was not uniform. It was speckled with dark spots. The black spots on the images were like flies walking across a television set. They were annoying. They were there from the start, on the very first images. Strictly speaking, these spots were areas of greatly reduced brightness. In other words, there seemed to be holes in the dayglow. There is no question about who saw them first. Everybody saw them. We would give talks and the black spots were there on the images for everyone to see. And everyone assumed they were noise--those random fluctuations in data that are due to chance.

Life went on. During the summer of 1982, an undergraduate student named John Sigwarth began working for me. Sigwarth had come to the University of Iowa in 1979 and had taken his first physics class with me. He was very interested in space science and was very bright. I wanted him to take enlargements of the satellite images and scan them for signs of gravity waves, small scale ripples in the upper atmosphere that sometimes follow the brightening of auroras. These waves sweep across the face of Earth much like waves in the ocean. They have been detected on radar and I wanted to know whether these waves would show up as crests and valleys of light in the satellite images. I assigned to Sigwarth the task of processing the data to make the waves visible.

But he could not do it. He worked on the computer program, which was designed to extract subtle features from the original pictures and highlight them, but each time the computer stuttered and burped as it scanned those black holes in the atmosphere. The black spots kept getting in the way. Then one day in the fall of 1982 Sigwarth, quite exasperated, walked into my office on the second floor of Van Allen Hall and said: "How can we get rid of this stuff?"

Sigwarth, Craven, and I pondered the question and concluded there must be some disturbance in the camera's electronics that would occasionally put these annoying little black spots in the image. Perhaps a transistor was fluky, or a computer was not functioning properly, or a problem arose as the pictures were being transmitted down to us from the satellite. It was possible. Each picture involves the transfer of an enormous amount of information. Like the image produced by a fax machine or by the display on a home computer, each of the satellite pictures is constructed of a large number of little dots called pixels. Each of these pixels is described by eight "bits," actually a string of "0's" and "1's" that indicate the intensity or brightness of that pixel. So perhaps some of the image pixels were not being transmitted down properly from the spacecraft. Whatever the source of the noise, I did not relish spending my time tracking it down. Sigwarth got the job.

It was tempting to simply remove the spots from the images and get on with the search for gravity waves. But you cannot alter data on a mere assumption. You have to have a reason. We needed to show that the spots were either detector noise, or produced by electronics on the spacecraft, or generated by computers on the ground. Only once that was accomplished could we eliminate the spots from the processed images and get on with our work.

Sigwarth worked very hard trying to solve the mystery. From time to time he would come into my office and say he was not having much luck with it.The year 1982 drew to a close. Sigwarth kept working on it. But he was unable to trace how the holes appeared on the images. One possibility that we entertained was that one of the two light counters on the camera was failing. Every other pixel in the satellite images is produced by an entirely separate set of electronics. The counters, in other words, take turns producing the dots that comprise each image.So Sigwarth had separated the data produced by each counter. But when he examined the data he saw that both counters were observing spots in the same sequence and at the same rate. This told us the counters were not dropping bits, those meaningful strings of "0's" and "l's." We knew that it was nearly impossible for two counters to malfunction in exactly the same way.

We also eliminated the possibility that these annoying little blackspots were caused by errors in radio transmission from the spacecraft. We checked the entire system from the time the data left the instrument, passed through the satellite itself, traveled down to the ground, and was relayed to us. Because the instrument regularly transmits fixed words, or fixed bit patterns, we could check to see if any transmission errors were occurring. We calculated that dark spots due to telemetry noise would appear once in every 200 images. But these spots appeared in the images almost a thousand times more often than expected.

All of the tests we performed on the data produced results that ran counter to our expectations. We really wanted to show that the holes were noise.But we were out of luck. So Sigwarth began to look at the pictures in great detail,trying to see if a spot that appeared in one frame could be seen in any of the subsequent frames. A very careful analysis showed that the spots could be followed in this way,although the spots in the subsequent exposures were not as dark as those in the initial exposures. This seemed to indicate that the black spots were moving and changing.

Sigwarth then programmed the camera so that instead of scanning the entire Earth, it would scan just a small portion of it. This allowed the camera to return to the same area more quickly. The series of pictures produced showed that a black spot would appear and disappear in a sequence of frames. The black spots seemed to be objects in motion. This was not characteristic of noise. Noise should appear at random all over the image. This indicated the presence of a real object.Sigwarth came down to the lab where I was working and showed me the data. He was very excited. I looked down at the pictures and congratulated him. I thought we were on to something.

Other clues convinced us that the spots were genuine. The spots,or holes in the atmosphere, appeared to move in the same direction across the face of Earth. If these holes were random events, due to malfunctioning equipment,for instance, you would expect to find half the spots going in one direction, and half going in the other direction. But this was not the case. Most of the holes appeared to move in the same direction across the face of Earth.

By February of 1983 we had come to the conclusion that something,some kind of object, was absorbing the ultraviolet radiation between the camera on the satellite and Earth and producing the apparent holes in the atmosphere. The more we looked the more it seemed that our images were actually snapshots of the movement of these objects above the atmosphere. We began to suspect that these objects were meteors of some unusual sort.

So we decided to compare the motion of our elusive black holes to the passage of meteoric dust and debris in our atmosphere. Much of this meteoric dust tends to orbit the Sun more rapidly than Earth. It essentially catches up to Earth, in other words, and approaches the atmosphere from the local evening face of the planet. Such motion relative to Earth is called prograde motion.Whatever the black spots represented, they showed the prograde motion that is characteristic of meteoric material. This not only implied that the spots were real but that the objects they represented were extraterrestrial. So we assumed, given the large number of spots that showed up in our images, that the holes were caused by a particularly large influx of meteors such as you would have with a meteor shower.

We went public with these results for the first time in May of 1983. Two weeks after receiving his undergraduate degree in physics, Sigwarth presented our findings in a paper entitled "Atmospheric Holes Possibly Associated with Meteors" at the spring meeting of the American Geophysical Union in Baltimore.I sponsored the paper and John Craven was listed as a co-author. People seemed interested and curious, nothing more. But over the next two-and-a-half years we presented three more papers on the topic at meetings of the American Geophysical Union and each time the audience grew.

Over this period of time Sigwarth and I analyzed over 10,000 images and learned a good deal about the black spots in the process. Our interpretation of the events continued to involve meteor impacts into Earth's upper atmosphere.By counting the spots in our images we were able to estimate the rate at which these objects appeared. This was the simplest measurement to do. We saw ten holes per minute on the daylight side of Earth. So we doubled that figure to obtain the rate of these objects over the entire face of Earth. There had to be about twenty such objects entering the atmosphere every minute. That was an alarming number of objects.

We still needed to explain just how these objects, which we assumed to be meteors, could cause holes in the atmosphere's screen of atomic oxygen. We entertained three possibilities. The first and simplest explanation had the meteors laying a blanket of material over the atmosphere, preventing the light from getting through, and creating a black spot in our images. Another possibility was that the atomic oxygen up there was being depleted by some special chemical process. But we could think of no chemical reaction that could get rid of the atomic oxygen quickly enough and none that allowed the atmosphere to restore itself as rapidly as we observed in our images. A third possibility involved a catalytic reaction of the sort that takes place in the catalytic converter in your car. Could some small amount of material,some catalyst, be converting the atomic oxygen in the atmosphere into molecular oxygen and producing the dark spots in our images? It was not likely, as we were never able to identify any such catalytic agent.

The knowledge that the spots actually moved across the face of Earth strongly pointed to the existence of some kind of an object that prevented light from passing through it. Whatever it was had to be big and blackening out the ultraviolet light at a certain wavelength. It could not be an atom. It could not be a rock. It could not be anything thrown up there from down here. It had to bea common molecule in the solar system that absorbs at the right wavelength. The only common molecule is water and water just happens to absorb at the wavelengths we were observing with our camera. There was no reason to look for anything exotic. Water,in the form of water vapor, fit the bill perfectly.

This explanation posed certain difficulties, all of which were more psychological than physical. When we calculated how much water we would need up there to produce a spot in our images, we came up with a figure of about a hundred tons. Anyone would tend to back off from such a large figure and initially we did too. Then we figured out how many such objects we needed to account for the holes in the images we observed over the course of the year. And it was not one, not a hundred, but ten million. There was the problem. One per year would not have been a problem. But ten million per year? Unfortunately, there was not much leeway in our numbers.

The size of the holes presented another problem. They were easy enough to measure. We knew the size of the area each pixel covered in our pictures and we knew the altitude of the spacecraft. But what looked like little dark spots on the images turned out, in reality, to be about thirty miles across. They could not be rocks because such large rocks would just smash the surface of Earth to pieces. These were clouds of water vapor.

It was the only reasonable explanation we could find. It was the only reasonable explanation that anybody has ever been able to offer. Early on some engineers suggested that the black spots might be due to satellite parts falling into the atmosphere. But that is not possible. First of all there are not ten million satellite parts out there falling down on us each year. And secondly, they are not thirty miles across. Many people said there had to be other explanations for the dark spots. But no one has ever come out with one.

The numbers were shocking. Earlier we had found that the spots on our images varied with the frequency of meteors falling into Earth. At first,we were elated. But when we sat down to think about it, we saw a disaster pending.The objects were just too big and too numerous to be just another nice little geophysical fact with no real impact on our thinking. But once we ruled out noise and ascertained that the spots were real, the next step, the interpretation, was trivial. The only way to interpret these events was in terms of ten million objects falling into Earth's atmosphere every year. That is an infall of material that is about ten thousand times more than anyone had ever imagined. So psychologically, emotionally, the interpretation was difficult to accept. But intellectually, it was trivial. There was no other reasonable explanation.

By December of 1985 the press had picked up the story. Stephie Weisburd had written an article for Science News entitled "Atmospheric Footprints of Icy Meteors" following our fourth presentation on the topic, which took place at the fall meeting of the American Geophysical Union in San Francisco.We still called the objects meteors at that time, but that was not quite right. One thinks of a meteor as dust and rock, not water, although some meteors are thought to contain a substantial amount of water before hitting the atmosphere. But the objects we were talking about clearly had to be mostly water. That could only mean one thing.Sigwarth and I finally realized after the meeting that these objects could be nothing other than comets, small comets.

People at the meeting seemed to be quite interested in our presentation but we were careful not to threaten anyone's perception of the solar system. We never came out and said just how big these objects were or exactly how many were falling into Earth's atmosphere. We just said there were a large number of unknown objects falling into Earth's atmosphere that had not been detected before. I also began discussing our findings privately, in other meetings and hallways. No one got too upset over it.

Science is no stranger to odd data. Sometimes anomalous observations are just put aside for later examination. But often they are just stuffed into a drawer and forgotten. We scientists normally have our hands full examining the data we were seeking in the first place. So odd data has a tendency to fall through the cracks. There is a lot of tucking under in science. Everyone thought I would eventually bury my data on the mysterious black spots. No one thought I would publish them.

They were wrong. I began writing the first draft of the original small comet papers in December of 1985. Eventually I handed out a few copies to people for their comments, but not before spending many nights pacing the hallways of Van Allen Hall and pondering the consequences of such an action. The papers touched on many of the objections that such a large infall of comets would raise. What would their existence mean for Earth and Venus and Mars and the other planets? What was the lifetime of such objects? What kind of mantle did they need to survive repeated passages through the inner solar system? Where did they come from?

One of the copies of our original papers had gone to Thomas Donahue,a professor of atmospheric sciences at the University of Michigan. You cannot do any better than Donahue when it comes to experts on planetary atmospheres. I respected his opinion. Donahue looked at our papers and said that if we were right these comets would have brought in enough water, over the age of Earth, to produce the oceans.And of course he was right. I had not thought of it because I had always assumed that the oceans had always simply been here. So did everyone else.

I submitted my first papers on the small comets to Alex Dessler in February of 1986. Dessler had just become editor of the American Geophysical Union's Geophysical Research Letters. He was ambitious and wanted to make the journal controversial, exciting and well-known. Unlike the quarterly Reviews of Geophysics, which Dessler had edited previously, this monthly publication provided a rapid way to communicate the latest research in the field of geophysics. Shortly after seeing our papers Dessler called to say that this was just the kind of material he wanted.He would have the papers reviewed in forty-eight hours.

Scientific journals rely on a rigorous review process to weed out inaccurate claims and research findings. Editors, if they are favorably disposed to manuscripts submitted for publication, send them to two or more scientists considered experts on the topic treated in the manuscript. The comments of these referees are returned to the editor, who then sends them on anonymously to the original author.This allows the referees to be candid and honest. Often, however, this power is abused.Those who are judging the merits of a manuscript are often in competition with the author for grants or recognition. Some scientists abhor this anonymity and those who know one another occasionally send their comments directly to the author of the submitted paper.

I had this kind of unspoken understanding with Donahue. He was one of the referees and he nearly exploded. He asked me not to publish the papers,the interpretation paper, in particular. The first paper was simply a description of the black spots and no one could deny that the black spots were there. The second paper was the interpretation. In it we spelled out that we were dealing with ten million, comet-like objects entering Earth's atmosphere per year, each one the size and weight of a small house. We also touched upon many of the topics this interpretation would seem to contradict, such as the origin of the oceans, as well as the lack of water in Earth's upper atmosphere, on Mars, on Venus and on the Moon, as much as we could cram, in other words, into four pages. This was the limit on the length of papers published in Geophysical Research Letters.

The other referee also recommended against publication. He said,Dessler told me, that "if this was correct, we would have to burn half the contents of the libraries in the physical sciences." It was a dicey situation for Dessler.The two people he had asked to review our papers had advised him not to publish.So Dessler called me and asked if I would withdraw the interpretation. He warned that its publication might destroy my scientific career. But I told him it was necessary.Without it people were likely to regard the black spots as a mere curiosity.

Donahue also begged me to come to my senses. He did not want the interpretation paper published. I think he was trying to be helpful. But he must also have been worried about the tremendous repercussions these findings would have for many fields of science, including planetary atmospheres, Donahue's own little niche. Yet despite two negative reviews and the "uncomfortable ramifications for the community at large," Dessler decided to go ahead with publication. The final decision always rests with the editor. Reviews are designed to guide, not bind.

"If you restrict the journal to publishing only what pleases the referees you end up publishing what is popular," Dessler explained in 1987." And while it does make everyone feel more comfortable, you are guaranteed to miss the occasional breakthrough. There is a price to pay, of course, for doing it the way I think it should be done. Nine out of ten of these things will turn out to be rather worthless. But I think the price is worth it. Occasionally one is so important that it makes publishing all the previous ideas that didn't turn out quite right worthwhile."

Anytime you break new ground in science you get attacked. This is the basic conservatism of science. But it is necessary to put all new ideas on trial. I had some notion of what the response to these papers would be. I had been involved in many scientific debates before. But the reaction I received was like none I ever experienced. I was driving a bulldozer through dozens of the neatly planted fields of science and everyone was upset.

I had, of course, been tempted not to publish these results. I was enjoying my work. People respected me. I had all the opportunities in the world and all the funding I needed. I did not need the aggravation these findings would stir up. On the other hand, I thought of all the scientists from Copernicus to Einstein who had come forward with seemingly outlandish but revolutionary results. I realized that if I was right and did not publish my results--and no one did this work for another ten or twenty years--then I would be wasting thousands of man-years of scientific effort.

People had warned me repeatedly and I understood what they were saying. Had I heard someone talking the way I did, I would have warned them as well.But I had agonized over the decision to publish this material for a long time. The decision came long after it became quite apparent that our efforts to show that the black spots were not real had failed. There were no other tests to be done. I knew this would have a big impact on science. But I felt I had no choice. I could not bury this material.

The notion of so much water falling in from space flew against all current observations and beliefs. Most likely the notion was wrong, so I had to make certain that it was not. I had done everything I possibly could to find something that said these objects could not exist. I only needed one big piece of evidence.Just one. A hundred pieces of evidence would not prove that they exist. But it would only take one to show that they did not. So I sat in libraries and read about astronomy, about oceanography, about geology, one field after another. And finally I decided that our findings must be published, no matter what the consequences. I could not live with myself otherwise. It was just morally incorrect.

The first small comet papers appeared in the April 1986 issue of Geophysical Research Letters, and are available in Adobe PDF format (as are the 10 REPLY papers written by Frank et. al. to the ten COMMENTS that were published by Geophysical Research Letters). These papers are Copyright ©1986, 1987 by the American Geophysical Union. Further electronic distribution is not allowed.