A sponge diver's chance discovery

The Mechanism is named after the Greek island of Antikythera [antih-KITH-ehra], where in 1900 a sponge diver taking shelter from a storm found an ancient shipwreck 200 feet below the Mediterranean's surface. Archeologists removed an array of artifacts, but it wasn't until mid-1902, that one of them noticed that embedded in what was thought to be a lump of broken calcified bronze statuary was a gear wheel.

The lump turned out to be an encrusted, partially destroyed set of bronze gear wheels, dials and plates. About 13 inches high, 7 inches wide and just under 4 inches deep, the mechanism was held in place by the remains of its original, well-made wooden case, which dried out and crumbled away after being excavated. Inscribed Greek letters and some inscriptions were visible on the metal plates.

Even then, the complexity of the Mechanism was obvious, so obvious in fact that some observers speculated that it actually was created during the Renaissance in the 1400s, had been lost overboard, and landed on the much more ancient wreck. Today, there is no question the Mechanism dates from around 65 B.C.

Study, speculation, theories

In 1974, a paper by science historian Derek de Solla Price and Greek nuclear physicist Charalampos Karakalos, based on gamma-ray and X-ray analysis, presented new details of the device. Price described the mechanism as a calendar computer and famously identified the presence of a differential gear -- a set of gears that can move two axles at different speeds.

In the late 90s, another research effort , including 700 X-ray plates digitized for computer-aided analysis, was carried out by Michael Wright, a specialist in the history of mechanicism and now with the Imperial Museum in London, and the late Allen Bromley of the University of Sydney.

In a paper presented a year ago, "Understanding the Antikythera Mechanism," Wright faulted some of Price's conclusions. Wright speculated that the Mechanism could be an elaborate planetarium, designed to show the movements of the five then-known planets in addition to the sun and moon, and to predict eclipses. He also suggested that the Mechanism was actually a marriage of at least two separate devices, one being added to the device already in the wooden case. He demonstrated that Price's differential gear was actually an epicyclical gear, in which a central wheel meshes with one or more peripheral gears, which rotate around the center.

Cardiff's Edmunds says the current research has not, in fact, found evidence of planetary displays, based on what's now known about the gear trains, but there is evidence of "planetary functions." He prefers the term calculator to computer: "It multiplies, divides and subtracts, but you can't program it," he says.

"Our interpretation will be somewhat different from both Price and Wright, though it will have some common elements," he says.

That interpretation has been made possible by advanced digital imaging and X-ray technologies.

Re-imaging the surface

Tom Malzbender, a senior research scientist with HP Laboratories, and colleagues Dan Gelb and Hans Wolters developed a digital technique, called reflectance imaging, for re-imaging how light is reflected from a surface. "By changing these reflectance functions, we could enhance the surface, and bring out details that weren't visible before," he says.

Essentially, it's a computerized version of what most of us have done with the oil dip stick in our car: you hold it up to the light and twist and turn it, until the light shows up the oil film and the inscribed markings. The HP researchers do it by putting an object inside a dome that's fitted with a camera, scores of light bulbs, and a laptop computer to control it. A separate laptop runs a program to create a polynomial texture map (PTM) of the captured images, letting the researchers then change the lighting and surface characteristics.

"If you make the [object's] surface 'like' obsidian, which has virtually no diffuse reflectants but has very strong specular [shiny highlight] characteristics, this brings out a lot of surface detail," Malzbender says.

Malzbender's work was presented in a 2001 SIGGRAPH paper that came to the attention of one of Edmunds' colleagues on the Antikythera project, mathematician and filmmaker Tony Freeth. Last fall, Malzbender and Gelb and Bill Ambrisco of Foxhollow Technology, traveled to Athens for a week to take photos of the front and back of each of the Mechanism's 70 large fragments, working 12 hours a day and more in a dark, cramped basement room with poor ventilation.

"It was completely exciting," he says. "We'd capture some pictures, and five minutes later we have the PTMs on the laptop and all these guys huddled around it, trying to read it." The HP crew captured and stored 4,500 images.