Welcome to Scribd, the world's digital library. Read, publish, and share books and documents. See more
Standard view
Full view
of .
Look up keyword
Like this
0 of .
Results for:
No results containing your search query
P. 1
M. G. Edmunds and T. Freeth (2011) The Use of Computation in the Decoding of the First Known Computer

M. G. Edmunds and T. Freeth (2011) The Use of Computation in the Decoding of the First Known Computer

Ratings: (0)|Views: 399|Likes:
Published by ArchaeoinAction
Using Computation to Decode the First Known Computer
Mike G. Edmunds, Cardiff University Tony Freeth, Antikythera Mechanism Research Project
Researchers have used many different kinds of software to analyze the structure and astronomical functions of the Antikythera mechanism’s surviving fragments. This ancient Greek calculator contains 30 gear wheels and has an extraordinarily sophisticated mechanical design.
Using Computation to Decode the First Known Computer
Mike G. Edmunds, Cardiff University Tony Freeth, Antikythera Mechanism Research Project
Researchers have used many different kinds of software to analyze the structure and astronomical functions of the Antikythera mechanism’s surviving fragments. This ancient Greek calculator contains 30 gear wheels and has an extraordinarily sophisticated mechanical design.

More info:

Categories:Types, Research, Science
Published by: ArchaeoinAction on Jan 21, 2012
Copyright:Attribution Non-commercial


Read on Scribd mobile: iPhone, iPad and Android.
download as PDF, TXT or read online from Scribd
See more
See less





Published by the IEEE Computer Society
0018-9162/11/$26.00 © 2011 IEEE
between 80 and 60 BCE; its rich contents are currently ondisplay at the National Archaeological Museum in Athens.One artifact from the wreck—a lump of corrodedbronze—split open after a few months in the museum,revealing toothed bronze gear wheels. Prior to this discov-ery, no evidence of metal gear technology in the classicalworld existed. Subsequent examination, fragmentation,and cleaning revealed an interior structure and inscrip-tions in ancient Greek that clearly indicated the device hadan astronomical connection and dated back to the secondcentury BCE. German philologist Albert Rhem was the firstto recognize, in 1905, that the mechanism was essentiallyan astronomical calculator. Its true complexity—30 gearwheels in situ—became apparent in the 1970s, throughthe use of photographic radiography by CharalambosKarakalos and Derek de Solla Price.
Subsequent work,most notably by Michael Wright,
added more mechani-cal understanding.In 2005, an international team of scientists initiated theAntikythera Mechanism Research Project (AMRP; www.antikythera-mechanism.gr) with the aim of obtaining freshdata by applying the latest imaging and analysis techniquesto the mechanism’s surviving fragments. So far, our inter-national team’s efforts have yielded both detailed surfaceimaging and microfocus X-ray computed tomography—the latter providing a complete 3D “body scan”—of thefragments. These efforts have led to the discovery of newfeatures, prompting a reinterpretation of the device’s over-all structure, and a mass of new inscriptions to decipher.
t seems appropriate that scientists are using moderncomputing methods to investigate the earliest knownmechanical “computer,” the Antikythera mechanism,made in ancient Greece. Technically, the mechanismis more of a specialized astronomical calculator or dis-play device, but the sophistication of its design is quiteextraordinary. So far, researchers have used an array of computational tools—simple spreadsheets, image analy-sis, simulations, and advanced animations—to decipherthe mechanism.
After more than 100 years of study, thecalculator’s functions—if not its ultimate purpose—are atlast reasonably well understood.The story of its discovery is widely known.
In 1900,sponge divers discovered the wreck of an ancient tradingship off the island of Antikythera in the Mediterranean Sea.From 1900 to 1901, the National Archaeological Museumin Athens and the Greek Navy carried out what was effec-tively the first major underwater archaeological recoverymission. Historians have dated the wreck to somewhere
Researchers have used many differentkinds of software to analyze the structureand astronomical functions of the Antiky-thera mechanism’s surviving fragments.This ancient Greek calculator contains 30gear wheels and has an extraordinarily so-phisticated mechanical design.
Mike G. Edmunds,
Cardiff University 
 Tony Freeth,
Antikythera Mechanism Research Project 
Using Computationto Decode the FirstKnown Computer
 JuLY 2011
On archaeological digs, it is standardpractice to take aerial photographs of thesite shortly after sunrise or just beforesunset when the sun is low in the sky. Theshadows cast by the glancing sunlightenhance faint surface irregularities ortexture differences on the land, makingunderlying physical structures visible.Tom Malzbender and his team atHewlett-Packard Labs in Palo Alto, Cali-fornia, developed a portable relightingtool for artifact surfaces that enabled usto obtain similar images of the Antiky-thera mechanism fragments. We placeda 1-meter-diameter geodesic hemisphereover each fragment, with 50 electronicflashbulbs distributed across the hemi-sphere’s inner surface, fired off insequence by a laptop computer that alsocontrols a digital camera at the dome’szenith, which takes a picture for eachflash. We use software to combine the se-quence of 50 differently lit images at will,with or without further image process-ing. When displayed interactively, theeffect is of being able to hold the objectin your hand and turn it in all directionsrelative to the light, greatly aiding thecomprehension of faint detail.Malzbender’s approach uses
 polynomial texture maps
 (PTMs; www.hpl.hp.com/research/ptm/ri.html) to rep-resent image pixels as functions of the lighting source’sdirection to specify red, green, and blue on surface com-ponents. A PTM fitter fits a low-order polynomial to lightedsamples of the object represented in the image sequence.We use a PTM viewer to evaluate this polynomial for eachpixel to produce a
reflectance transformation image
(RTI)that enhances the object’s surface detail under variablelighting conditions. Even low-end computers can handlethis reconstruction at real-time rates because of the poly-nomial’s simplicity.Figure 1 compares a photo and RTI of fragment 19, themechanism’s back cover. Interactive RTIs of all 82 surviv-ing fragments are available at www.hpl.hp.com/research/ptm/antikythera_mechanism/index.html. Examples of awide range of PTM applications to other archaeologicalartifacts such as stone tools, ceramics, coins, and rock artcan be viewed at http://c-h-i.org/examples/ptm/ptm.html.
In archaeology, computed tomography (CT) involvesmounting an object of interest on a turntable and takinga series of 2D X-ray projection images called radiographsvia computer as the turntable rotates 360 degrees in an-gular steps. Researchers analyze the resulting data withspecial software that reconstructs the 2D projections intoa 3D volume.X-Tek Systems (now part of Nikon Metrology) built andoperated for us a special prototype 450-keV BladeRunnerCT system, the high energy of which could penetrate theAntikythera mechanism’s largest surviving fragment. Thesystem’s
X-ray source has a small but intensebeam diameter that allows much greater spatial resolutionthan typical CT medical scanners. The detector was a 16-inch Perkin Elmer flat panel with 2,048 × 2,048 square400-micron pixels.With this microfocus source, we could geometricallymagnify fragment samples on the detector by manymagnitudes, allowing resolutions in the range of 40 to100 microns, depending on sample size. We imagedmost of the fragments at X-ray potentials of 225 keV and366 keV, scaling the 16-bit monochromatic radiographsfrom the detector to fit the dynamic range by convert-ing them to 8-bit images with 256 gray levels, beforecompressing and saving them as JPEGs. Example imagesof mechanism fragments can be found at www.shawin-spectionsystems.com/library/antikythera/dr/dr.htm.
Figure 1.
Applying surface imaging to fragment 19, the Antikythera mecha-nism’s back cover: (left) photograph and (right) reflectance transformationimage that enhances the text in what amounts to the mechanism’s instructionmanual. The text highlighted in red identifies the ancient Babylonian cyclesthat underlie the gearing: “76 years” refers to the Callippic cycle, “19 years” tothe Metonic cycle, and “223” to the Saros cycle.
Copyright 2006-2008 Antiky-thera Mechanism Research Project 
The key analysis tool for the X-ray CT was VolumeGraphics’ VGStudio Max, a powerful software programfor analyzing 3D X-ray volumes that runs on a standardPC or Mac. For our volumes, we needed a minimum of 8 Gbytes of RAM on a machine running four processorcores, and we could have used considerably more com-puting power. VGStudio Max produces 3D images, but, asFigure 2 shows, we found that the most powerful techniquewas to view slices of each fragment through the recon-structed X-ray volumes.These slices can be angled at any orientation, which isfortunate because nothing in the surviving fragments istruly flat. The software also includes sophisticated mea-surement and shape-fitting tools to extract metric data forconstructing a model.We anticipated that the X-ray data would help us try tounderstand the 3D disposition of the mechanism’s gears,but one of the surprising revelations was that we could alsouse it to read many new inscriptions hidden deep insidethe fragments. Our Greek colleagues in the AMRP, YanisBitsakis and Agamemnon Tselikas, took the lead on deci-phering these inscriptions. They found more than 2,000additional text characters, some via RTIs but the majoritythrough X-ray CT. Most of the new text could not be read ina single CT slice—for one of the calendar dials, we neededmore than 60 slices, spaced at 100 microns apart, to readall the month names. The names, as interpreted by ourcolleague Alexander Jones, imply that the machine wasdesigned for use in Corinthian Greece.
After we obtained the X-ray data, we next neededto establish reliable tooth counts for all the gears. Themechanism’s astronomical function is essentially en-coded in the gear ratios, although the inscriptions onthe mechanism do give some valuable hints. We believethe gears were laid out and cut by hand, with the teethappearing to be simple equilateral triangles, slightlyrounded at the tips. Only five of the surviving 30 gearwheels have a complete set of teeth, although withsome irregularity in tooth spacing; the others also showvarying amounts of damage, incompleteness, and physi-cal tooth spacing. For each gear, we imported severalparallel CT slices into a CAD program, Nemetschek’sVectorworks, which let us superimpose a geometry onthe slices. We could then estimate the assumed gearcenter and mark the angular positions of the surviv-ing gear teeth tips. Next, we exported data to an Excelspreadsheet for a tooth-count analysis.Based on our examination of the gear image, we se-lected potential contiguous runs of teeth and read themfrom the spreadsheet into a Mathematica program writ-ten to fit a model with
perfectly spaced teeth. This gaveus an arbitrary start position and allowed us to move theassumed gear center around. The “goodness of fit” pa-rameter was simply the reciprocal of the least-squaresdeviation between model and data, but it worked remark-ably well. In most cases, a well-defined maximum inthe fit parameter appeared when plotted against
Figure 2.
Typical X-ray CT slices, in a false color scale. (a) A gear with 127 teeth in the center that calculates the mean motion of the moon against the stars, according to the Metonic cycle. The large partial gear originally had 223 teeth and drives theSaros dial. (b) The 127-tooth gear, marked in Vectorworks for counting the teeth.
Copyright 2006-2008 Antikythera MechanismResearch Project 

Activity (4)

You've already reviewed this. Edit your review.
1 thousand reads
1 hundred reads
Ravi Kn Ravi liked this
windua liked this

You're Reading a Free Preview

/*********** DO NOT ALTER ANYTHING BELOW THIS LINE ! ************/ var s_code=s.t();if(s_code)document.write(s_code)//-->