Gravitational Radiation

According to electromagnetic theory, there should be electromagnetic radiation, and there is: examples include light, radio waves, microwaves, ... .

Einstein's theory of general relativity in many mathematical respects looks like the theory of electromagnetic radiation (Maxwell). That is no accident: Maxwell's theory was always Einstein's favorite theory.

One might therefore suspect that the theory of general relativity would predict that there is gravitational radiation, and indeed it does. This was viewed as very exciting for several reasons. When I went to college, one of the first physics courses I took was a course on the theory of gravitational radiation.

The idea of gravitational radiation is pretty routine compared to a lot of general relativity theory. Consider, for example, black holes. In fact, it is now generally accepted that there are a lot of black holes: one in the center of most galaxies.

It is "easy" to generate gravitational radiation: any rapid vibration of mass should do it. Thus, given how confident we are about general relativity, it is a safe bet that there are gravitational waves. Thus, the serious question is not whether there is gravitational radiation, but whether we've detected it. If we keep trying and keep failing, that will be a serious problem for our confidence in general relativity and various ideas in astrophysics.

Gravity is a lot weaker than electromagnetism, and so gravitational radiation is very weak. We also don't know much about how or where it might be generated, at what frequencies, how continuously, and so forth.

Unlike, say, the Michelson-Morley experiment, we have no clue what power, what frequency, what direction to look.

Weber built the first gravity wave detectors. He detected something. In order to have any chance at all, he tuned it to only one frequency. He reported detecting a lot of gravitational radiation. In fact, he reported detecting so much that it was a substantial portion of all the energy in the universe. If there was really that much gravitational energy all around all the time, the universe is going to flame out soon.

How do you detect gravitational radiation? Gravitational radiation causes all the gravity it goes through to fluctuate, and so the rate at which things fall and attract each other should jiggle a bit. The changes are tiny. What Weber, and all experimenters since, have done is isolated a large massive object, and tried to detect a jiggle. He used an aluminum cylinder in a vacuum suspended from a string supported by rubber and lead vibration dampers connected to a phonograph needle. He recorded the output from the needle. Now, people use huge cylinders and try to detect motion using SQUIDS, which use quantum effects to measure motion.

What you get from any of these devices is a huge amount of noise. The noise is much bigger than any possible effect. So, how can you be sure you've found a signal in all the noise? One does statistical analysis, one looks for peaks of certain shapes, one sets up several detectors widely separated and looks for coincidences. You look for patterns that reappear each day when Earth is pointed toward the same place in interstellar space.

Weber got results that passed all of these tests. On the other hand, if you let monkeys loose on typewriters for long enough, they would eventually type out the complete works of Shakespeare.

So, the question is, do we believe Weber? What are standard scientific practices in evaluating an experiment? The standard answer is that you try to duplicate the results. Lots of people did, most failed. That doesn't really settle anything: each side simply concludes that the other is doing a bad job. What about the tests? Statistical significance, coincidence, daily period.

Everyone agrees that all of these are relevant, but there is no agreement about their relative importance in this case. Weber tried correlating his results with those of a rival. He found coincidences. Unfortunately, he screwed up about daylight savings time and found coincidences between signals at different times. That was very bad. He also reported results of a computer analysis that turned out to involve a programming error.

What about other labs? They didn't know whether they "should" detect radiation or not, and so they tended to be cagy, waiting for a consensus.

A very famous and very influential physicist, Dick Garwin, "decided to see if he could put a stop to this" and published a very negative review article and declaring that nothing had been detected. With Garwin as cover, everyone doing these experiments was suddenly willing to risk publishing negative results. That fixed the consensus.

Does that mean that cutting edge experiments provide no objective check on theory? Is it all entirely social and political? Collins and Pinch have often been accused of thinking so, but they do not, and it is not entirely social and political.

Two examples: Lysenkoism and global warming.

-- ShaughanLavine - 23 Nov 2005