Since the dawn of X-ray astronomy in the 1970s, astronomers have puzzled over why the powerful X-ray emission from double star systems known as X-ray binaries is so variable.

An X-ray binary system contains either a neutron star or a black hole. These bizarre objects are only the size of a large city, yet are more massive than our Sun -- and in X-rays alone, emit ten thousand times more power than the total output of the Sun.

This immense amount of energy is generated because material from a normal companion star flows onto the neutron star or black hole, attracted by its very strong gravity.

But the X-ray output isn't steady. It flickers very rapidly, varying by small amounts on very short time-scales (fractions of a second). At the same time, the X-ray output varies by larger amounts on longer time-scales.

No satisfactory explanation for this variability has ever been proved but a favorite idea among astronomers is called the "shot-noise model" (borrowing a term from electronics).

According to this theory, the variability is caused by random flares or "shots" occurring in the X-ray emitting region within the star system. Like building blocks piling up, the shots combine together to produce variations on all time-scales.

Because the individual shots in the shot-noise model occur randomly and are independent of all the other contributions to the total X-ray output, the amount of variability on short time-scales should always be independent of what happens to the X-ray output on longer time-scales.

But using data from observations of X-ray binaries made with NASA's Rossi X-ray Timing Explorer satellite, researchers at the University of Southampton Dr. Phil Uttley and Prof. Ian McHardy have discovered that just the opposite is true.

By comparing the amount of variability in many small segments of data with the overall X-ray output (or "flux") measured in each segment, they found that the amplitude of short-time-scale variations increases in direct proportion as the X-ray flux increases on longer time-scales.

In fact, they found a perfect linear relationship between X-ray variability and X-ray flux. This result implies that the short-time-scale variations in flux must somehow "know" what is happening to the flux on longer time-scales. This finding is totally inconsistent with standard shot-noise models.

However, there is an alternative theory that fits with the observations.

A number of theorists have noted that variations in the X-ray output of X-ray binaries might be caused by fluctuations in the flow of material that fuels the X-ray emission, taking place far outside the region of space where the X-rays are actually produced.

In this scenario, slow long-time-scale variations in the flow of gas from the normal star happen far from the black hole and move inwards like ripples on a pool of water. As they move inwards, the slow variations in the fuel supply pick up more rapid variations, which are larger when the fuel supply is larger, so that when the fueling rate -- and hence X-ray output -- is high, so is the degree of variability.

Dr. Uttley said, "With this discovery we have dealt a fatal blow to the most popular theory for why X-ray binaries are so variable. We are close to pinning down the real answer and resolving a mystery that has intrigued X-ray astronomers for 30 years."

One twist in the story is that this important property of X-ray binary systems could have been discovered in the early days of X-ray astronomy, had astronomers known what to look for.

In X-ray binaries, the pattern of increasing variability with flux cannot easily be seen by just looking at the data with the eye; computer analysis is needed to reveal it.

Dr. Uttley and Prof. McHardy were pointed to the existence of this pattern in X-ray binaries by their observations of the much slower variations in distant "active galaxies," which contain black holes a million times larger than those in X-ray binaries -- and which vary much more slowly.

In a few of these objects, the pattern of increasing variability with flux is easily recognizable in the data without the need for computer analysis, although a consequence of the slower variations is that the perfect linear nature of the relationship found for X-ray binaries cannot be discerned.

(Reference: Monthly Notices of the Royal Astronomical Society Vol. 323, No. 2, 11 May 2001.)

(Editor's Note: The full paper can be found on the web at this URL.)

10-May-2001