Bak and others had written, he asked himself, “Well, what would happen if we tried this in real life?” How often
would
those sand avalanches occur? Were the ideas in Bak’s head also true in the lab? Bak had first suggested the idea in a theoretical
journal; he didn’t seem to have much of an intention to test it in reality. Could it be done? How do you build a sandpile
grain by grain? No one had ever attempted it. So Glenn Held decided to give it a try.
To begin with, Held needed grains that were more or less the same size, since clumps of sand would distort the evenness of
the pile. In pursuit of something reliably sandlike, he began experimenting with grains of aluminum oxide, but he eventually
discovered that beach sand, collected on weekend trips to the shore, was more than sandlike enough. Held filtered what he
collected to get grains of more or less the same size. He made sure that the grains were all dry (any moisture would have
altered their weight and interfered with the experiment). Then he placed the sand in a jerry-rigged machine that looked sort
of like an automatic pepper mill connected to (of course) an IBM PC. The computer controlled how fast the mill turned and
how many grains trickled out on each rotation. This allowed Held to drop the grains precisely onto a palm-sized plate. He
put the plate on a scale so he could measure how much sand was falling on and off the pile (each grain weighed .0006 ounce)
and then propped the scale inside a Plexiglas case so stray air-conditioning breezes wouldn’t disturb his pile. Building the
device took him about ten hours. Then Held turned it on.
The first sandpile, which took a day to build, the grains dropping carefully one at a time, was about two inches across on
the bottom. Just as Bak had predicted, during an initial period the pile shaped itself into a cone, an example of what physicists
call “self-organization.” No one was telling the grains where to go; the intrinsic physics of falling sand just meant that,
over time, they sorted themselves into a nice even pile instead of spraying all over the place.
Once a pile reached a certain size, Held saw, it entered that strange “critical” state Bak had anticipated. Sometimes one
additional grain would cause an avalanche; other times Held could add thousands of grains before the sand started sliding
off. Held discovered a surface pattern to the avalanches, something scientists call a “power law,” which also applies to the
distribution of other nonlinear natural phenomena, such as earthquakes. (Charles F. Richter, the father of the Richter scale,
teased the pattern from centuries of earthquake data: large earthquakes occur exponentially less frequently than small ones.
This is called a “power-law distribution.”)
But the most interesting thing about the sandpile was its fundamental unpredictability. You couldn’t take your eyes off it
for an instant. The power law told you the general chances of getting an avalanche, but would that next grain of sand set
one off? Traditional science saw the sandpile as stable, in an equilibrium state, something that might have an avalanche when
disturbed by some strong outside force or when the number of grains reached a particular quantity. But Held’s sandpile behaved
quite differently. There was no magic number. One additional grain of sand was as likely to start an avalanche as a dozen.
What happened
within
the pile, the shifting and sliding of the grains, was as important as what happened
to
the pile. There was no explicit link between how you hit the pile and how it responded, no “proportionality” between cause
and effect. Just as Bak had theorized, the sandpile was a system that could “break down not only under the force of a mighty
blow, but also at the drop of a pin.”
Held wondered how you might model such a system. Frankly, this was a difficult puzzle. Every grain on the pile was, in a sense,
Amanda Hocking
Jody Lynn Nye
RL Edinger
Boris D. Schleinkofer
Selena Illyria
P. D. Stewart
Ed Ifkovic
Jennifer Blackstream
Ceci Giltenan
John Grisham