Einstein's Special Theory of Relativity states that information cannot travel faster than the speed of light in a vacuum. In some highly specialized "fast-light" media, however, some experimental physicists believe they have seen light travel faster.

Meanwhile, theoretical physicists have proposed that while smooth pulses of light may appear to travel faster in fast-light media, they don't carry information. It is the discontinuities in light signals that carry information and these discontinuities will travel no faster than light in a vacuum.

But very little experimental work has been done to support this theoretical analysis, and experimental physicists would like to see actual physical proof from the lab that supports the theory.

In this week's issue of "Nature" magazine, Mark A. Neifeld, a UA Electrical Engineering and Optical Sciences Professor, and Duke University physicists Michael D. Stenner and Daniel J. Gauthier present the experimental results that experimental physicists have sought and that confirm that Einstein does, in fact, continue to be right.

Their experiments show that information is indeed limited to speeds lower than C (Physicists use the letter "C" as a symbol for the speed of light in a vacuum.), and information that appears to travel faster is the result of reading too much into where information is carried on light pulses.

A smooth pulse doesn't carry information on its own, Neifeld explained. These smooth (Gaussian) pulses are called analytic signals. That means that if you look at any small piece of that signal, you can predict what the entire signal will look like for all time. If a signal has enough smoothness in it, it can't carry information, he noted.

The UA and Duke researchers devised an experiment in which smooth pulses of light (Gaussian pulses) propagated through a fast-light medium of pumped potassium vapor. These pulses appeared to travel faster than C. Then they added digital information to the pulses in the form of discontinuities. As the pulse got near its peak they either increased or decreased the amplitude, producing two states that could represent binary information (a one or zero, for instance).

The research team developed a way to describe the bit-error rate of these pulses as a function of the elapsed observation time. The researchers knew that information was being transmitted when they observed lower bit-error rates.

Prior to the arrival of a discontinuity the bit-error rate is 1/2 because no information has been received. After information starts arriving at the detector, the bit-error rate decreases with increasing observation time. A bit-error rate of 1/2 means there is an equal probability that the signal is carrying a one or zero and you might as well toss a coin to decide what was transmitted.

When the discontinuities were added to the Gaussian pulses, the researchers watched for the bit-error rate to fall below a threshold value that was less than 1/2. These observations allowed them to conclude that the information traveled at speeds slightly less than C.

Stenner and Gauthier at Duke conducted the experimental part of the work and Neifeld designed the encoding system that allowed the researchers to add information to the optical signals. He also helped define the types of detection techniques to use and showed that bit-error rate was the proper way to quantify the transmission of information.

So why do some light pulses appear to travel faster than C in fast-light media, but information always travels slower?

Neifeld explained that the problem comes in treating the peak of a smooth light pulse as an information carrier. Since it does not carry information (it has no discontinuities) it does not violate special relativity by going faster than C. This experiment has shown that the velocities of smooth pulses and of light pulses that carry information are distinct, he said.