PDA

View Full Version : a few good articals on the Antikythera



kendall
12-18-2008, 02:13 AM
the Antikythera is featured in a few stories this morning:


http://blog.wired.com/gadgets/2008/12/2000-year-old-a.html?npu=1&mbid=yhp

Other pages reached through links on that page, well worth a read.

keep wondering what the world would be like now if the knowledge and means for building that was in widespread use at the time. It's just a short little half step from something like that to automation as with Babbage, and not much farther at all to adding machines.

Ken

lazlo
12-18-2008, 08:42 AM
Charles Babbage's "Analytical Engine" was a true mechanical computer -- it read "programs" on punched cards.

Most researchers (including Michael Wright, the guy shown in the video) believe it was a geocentric ornery (astronomical clock). The Antikythera is 27 gears on a central axis, and the instructions engraved on the back say that you turn the knob (which is the only "input") once a month.

Michael Wright (from the London Science Museum) and Allan George Bromley made a working model from X-ray tomography. Wright demonstrated his model on the "Ancient Mysteries" show on the History Channel, and the model is now on display at the Antikythera museum in Athens.

I posted Micheal's paper here awhile ago which shows the original pieces, plus the X-rays, and a couple of pictures of the completed model (the one he shows in the video):

http://www3.imperial.ac.uk/pls/portallive/docs/1/24853696.PDF

Your Old Dog
12-18-2008, 09:20 AM
Thanks. I found the article most depressing. Here they are doing this kind of stuff at nearly the dawn of civilization and I'm still trying to learn how to thread and I got a shop full of equipment :D Maybe I need to scrap the machinery and outfit myself with some state of the art rocks.

Teenage_Machinist
12-18-2008, 04:05 PM
The antikythera device is simple, but such clockwork certainly could have progressed farther. I wonder what it would be like if the Greeks/Romans never fell, if there had been no Dark Ages, no Catholic Church, and no Feudalism.


We might have a god of computers-- Epithets could be weird.

Evan
12-18-2008, 04:29 PM
Robert, In one of the links they correctly call it a mechanical analog computer. The definition of an analog computer is not the same as for a digital computer. An analog computer computes something and is not necessarily a programmable device. A Norden bombsight is an analog computer as are the mechanical fire control computers used on ships in WWII.

An analog computer can be as simple as an operational amplifier that produces a signal that is the product of two inputs. It computes by the nature of it's design. So does the Antikythera mechanism.

It is interesting that it was used to compute the time of the upcoming Olympic games. That is in keeping with one of my guesses that it had a military purpose as the games were highly military oriented and the majority of competitors were soldiers.

lazlo
12-18-2008, 04:35 PM
Robert, In one of the links they correctly call it a mechanical analog computer.

It's media hype to call it an analog computer, IMHO. None of the scholarly articles refer to it as that. It's just a clock :)


An analog computer can be as simple as an operational amplifier that produces a signal that is the product of two inputs.

Sure, an op amp takes an arbitrary input signal, and multiplies it by some constant. Or an analog fire control computer takes position and wind direction as input, and calculates ballistics trajectories to hit it.

But the Antikythera has a single knob as input, and "all" it does it rotate the central gear spindle. The 27 gears are various ratios of the central gear chain -- it's not calculating anything.

Now, it's definitely a very cool clock/ornery...


It is interesting that it was used to compute the time of the upcoming Olympic games.

It doesn't "compute" the time of the upcoming Olympics, any more than your watch computes anything. The Olympics indicator is on a four year gear ratio. So if you turn the knob 12 months * 4 years = 48 times, the dial will point to "Olympics!" :)

Evan
12-18-2008, 04:50 PM
It still fits the definition of analog computer. An op amp multiplier doesn't even have one knob.



Analog Computer Museum


--------------------------------------------------------------------------------

Definitions
Computer - A machine that caculates.

Analog computer - A computer that performs mathematical operations in a parallel manner on continuous variables. The components of the computer are assembled to permit the computer to perform as a model, or in a manner analogous to some other physical system.

Mechanical analog computer - An analog computer with input and output usually expressed as shaft postion or degrees of shaft rotation.

Electronic analog computer - An analog computer with input and output usually expressed as direct current voltages.

http://dcoward.best.vwh.net/analog/analog1.htm

lazlo
12-18-2008, 05:49 PM
So by your definition, my gearhead lathe is an analog computer -- the output shaft rotates at a gear ratio of the input shaft.

Actually, my gearhead lathe is more advanced than the Antikythera mechanism, since I can select the ratio on my lathe, and in the Antikythera the ratios are fixed.

dp
12-18-2008, 07:18 PM
The Antikythera just like op amp and other analog computers are simply reporting devices. They report the state of their components. The operator changes the state of components in some way and the consequential result is presented.

Evan
12-18-2008, 07:36 PM
So by your definition, my gearhead lathe is an analog computer -- the output shaft rotates at a gear ratio of the input shaft.


If you use it to do ratiometric calculations by that means then it would be an analog computer. BTW, that is not my definition.



The Antikythera mechanism is believed to be the earliest known mechanical analog computer.[2] It was designed to calculate astronomical positions. It was discovered in 1901 in the Antikythera wreck off the Greek island of Antikythera, between Kythera and Crete, and has been dated to circa 100 BC. Devices of the level of complexity as the Antikythera mechanism would not reappear until a thousand years later.

The astrolabe was invented in the Hellenistic world in either the first or second centuries BCE and is often attributed to Hipparchus. A combination of the planisphere and dioptra, the astrolabe was effectively an analog computer capable of working out several different kinds of problems in spherical astronomy.

http://en.wikipedia.org/wiki/Analog_computer

John Stevenson
12-18-2008, 07:45 PM
It's an early quill feed for a Bridgeport, and in about the same condition......................

.

Evan
12-18-2008, 07:50 PM
Actually, my gearhead lathe is more advanced than the Antikythera mechanism, since I can select the ratio on my lathe, and in the Antikythera the ratios are fixed.


Not necessarily. According to the device website there is at least one adjustment for the moon orbital anomaly and there may be one for the sun as well.

lazlo
12-18-2008, 08:49 PM
If you use it to do ratiometric calculations by that means then it would be an analog computer. BTW, that is not my definition.


http://en.wikipedia.org/wiki/Analog_computer

If you want to quote Wikipedia, they call the Antikythera an orrery, and most people don't consider a orrery to be an analog computer:

http://en.wikipedia.org/wiki/Orrey

An orrery is a mechanical device that illustrates the relative positions and motions of the planets and moons in the solar system in heliocentric model. They are typically driven by a large clockwork mechanism with a globe representing the Sun at the centre, and with a planet at the end of each of the arms.

The Antikythera mechanism may be considered one of the first orreries.

By the way, the same schlock web site that composed that video, "New Scientist", also claims to have reconstructed an Ancient Greek programmable robot :D

In the interview, the author talks about how it uses string wrapped around the axle to "program" the robot, and how that's "just like modern programming languages." :p

Yep, a couple more pieces of string and the Ancient Greeks woulda had R2D2! :D

New Scientist recreates a robot made by the ancient Greeks
http://www.youtube.com/v/xyQIo9iS_z0&hl=en&fs=1"

saltmine
12-18-2008, 09:24 PM
Sometimes it don't take a whole shop full of machine tools, DRO's, and computers. Those little buggers in the black pajamas, in Viet Nam, could make duplicate parts for an AK-47 with nothing more than a worn out file, and a broken hacksaw blade. Same thing's happening over in Iraq and Afganistan...Those guys spend weeks scraping and filing on a piece of scrap metal (usually a blown up tank or truck) and fashion all sorts of parts...Gad! I'd hate to see what one of them could do if he had access to a well equipped machine shop....

Evan
12-18-2008, 09:57 PM
and most people don't consider a orrery to be an analog computer:


They do if they know what an analog computer is.

Robert,

Analog computers are not programmable in the sense that digital computers are. An analog computer is essentially a simulator of the problem to be solved. An orrery qualifies as an analog computer. Analog computers were the first type of computer I worked with and some implementations can be reprogrammed via various types of input devices such as switches, pots, resistor packs and may also take input from various sensor types. Or, they may simply perform a mechanical calculation to arrive at some sort of answer.

A slide rule is an analog computer and it only has a single movable part. Yet it was used to do the calculations to build the SR71.

lazlo
12-18-2008, 10:06 PM
Evan, as you well know, I'm a computer architect. I know quite well what an analog computer is :)

A slide rule is indeed an anolog computer. But a collection of gears with a fixed gear ratio on a common spindle with no input is not an analog computer, any more than some rope wrapped around a wooden spindle is an Ancient Greek programmable robot.

Like I mentioned earlier, the Antikythera is nearly identical to the gearset on a gear-head lathe.

There's a new diagram of the Antikythera gear arrangement in the new article this month in Nature -- I'll post it in a couple of minutes. It looks like something a HSM'er could make in a weekend.

lazlo
12-18-2008, 10:31 PM
Nature published a new article this past July on the Antikythera:

"Calendars with Olympiad display and eclipse prediction on the Antikythera Mechanism"

http://www.nature.com/nature/journal/v454/n7204/extref/nature07130-s1.pdf

It contains a very nice diagram of the internal mechanism -- you can decide for yourself what to call it:

The Earth is on the central shaft (the Greeks still believed the Earth was at the center of the universe), and the indicators for the Sun, Moon and the 5 known planets revolve around the central axis (the earth). Those are the gears from the centerline up, which are driven by the crown gear (the clock knob shown in the video) on the left.

The gears below the centerline are the Metonic and Callipic calendars displays:

From the Nature article:

Ancient calendars were based either on the synodic lunar cycle, the solar cycle or both. Twelve lunar months is about 11 days short of a year, so calendars on this basis do not remain synchronized with the seasons. Attempts to rectify this meant finding integer periods of years, which are also integer numbers of lunar months. One of the most accurate of these is the 19-year cycle of 235 lunar months, attributed to Meton of Athens in the 5th Century BC. The Metonic Cycle is one of the two basic cycles that underlie nearly all the known gearing of the Antikythera Mechanism.

This is crying-out for one of the Horologists here to make :)

http://i164.photobucket.com/albums/u15/rtgeorge_album/antithikyra.gif

lazlo
12-18-2008, 10:33 PM
And here's a great Java simulator of the Antikythera mechanism that I posted before that shows the relative motion of the 5 gears on the front and show the display changes when you turn the knob each month:

http://www.math.sunysb.edu/~tony/whatsnew/column/antikytheraI-0400/kyth5.html

ulav8r
12-19-2008, 12:22 AM
Well at least someone finally got the spelling right. I'm an ornery &^%$&%^$#^&%.

Evan
12-19-2008, 05:33 AM
This is crying-out for one of the Horologists here to make

I wouldn't attempt it, mainly because they haven't finished deciphering all the functions. As I said, there appear to be adjustments possible that alter the function. Even if there are not it meets the definition of analog computer. As a computer architect you should know that.



A device that receives, processes, and presents information. The two basic types of computers are analog and digital. Although generally not regarded as such, the most prevalent computer is the simple mechanical analog computer, in which gears, levers, ratchets, and pawls perform mathematical operations—for example, the speedometer and the watt-hour meter (used to measure accumulated electrical usage).

http://www.answers.com/topic/computer-1

dp
12-19-2008, 12:22 PM
I don't see this as receiving and processing information. That was built into the machine. Turning a crank that spins dials is no more a calculator than is a kid's wind-up toy that causes a monkey to dance and hold out its hat, or rapidly flipping through the pages of a 10,000 page calendar. The best I could accept is it is equivalent to a rotary decoder where the encoding was set by the builder. Nothing computational going on there. And the only reason it can be used as a decoder is because the operator already knows what to expect as a result: A full moon, for example, or an eclipse. Both events being available in a 10,000 page calendar should someone take the time to do the math in preparing the calendar as did the person who designed the antikythera.

Evan
12-21-2008, 01:36 PM
I don't see this as receiving and processing information

Turning the input knob is the information input. That information is the position of the knob as defined above. The information in question, I presume, is a date. Given that date information the device mechanically calculates what occurs on that date. The device is probably calibrated so that one of the dials indicates what date the knob is set to.

The user may expect an eclipse at some time in the future but the device calculates precisely when the eclipse occurs. The fact that the results are entirely deterministic makes it no different than any other form of computer as all computers are entirely deterministic. The fact that such determinism may be obscured by increased complexity does not alter the nature of a computational device. It is a difference of degree, not of kind.

Even a modern digital computer, given a particular input, is no more than a set of cogs and wheels and trip levers implemented as electronics that produce an entirely deterministic output given a particular input.

Behind the apparent complexity of a modern digital computer is ultimate simplicity. There are only two boolean functions needed to construct a computer of any arbitrary degree of complexity.

AND, which is "if A (is true) AND B (is true) then C (is true)"

NOT, Which is "B=NOT A" ( if A is true then B is false, If A is false then B is true )

That's it. While a clockwork assemblage of gears and wheels is visibly comprehensible to the eye a digital computer is not. That does not alter the basic nature of the machine.

A mechanical analog computer is visibly simple but it is no more simple except by degree than any other computer. The output is determined by the input as it is in all computers.

BobWarfield
12-21-2008, 02:01 PM
The Antikythera just like op amp and other analog computers are simply reporting devices. They report the state of their components. The operator changes the state of components in some way and the consequential result is presented.

dp has it right. Antikythera is just so, and the knob is not a wind up mechanism, but an input for date, relatively speaking.

And yes, a gearhead lathe is a fine if specialized analog computer as anyone who has tried to do the calculations by hand for threading will attest. It performs exactly the kind of specialized function most analogs do, and if a slide rule qualifies, both the antikythera and the lathe do.

You boys quit your bickering, it was boring years ago and accounts for a lot of the decline of this board.


Those little buggers in the black pajamas, in Viet Nam, could make duplicate parts for an AK-47 with nothing more than a worn out file, and a broken hacksaw blade.

Ever take an AK-47 (actually the legal semi-auto variant) apart? They are amazingly simple. The genius was in designing such a mechanism so that little tolerance was required for it to operate.


This is crying-out for one of the Horologists here to make

I quite agree! Evan, you have an interest in astronomy, forget how the original works, design your own orrery. It's a fascinating project, far more rewarding than online hair splitting. I have been fooling with it myself off and on since reading the original articles on antikythera. Here is my page of notes on the project:

http://www.cnccookbook.com/CCOrreryNotes.html

http://www.cnccookbook.com/img/Antikythera/ZodiacPerspective.jpg

If I can get my CNC technique working well enough, I may have a go at making one.

Cheers,

BW

dp
12-21-2008, 02:17 PM
Turning the input knob is the information input.

Then winding a clock spring is information input. I don't consider a clock to be an analog computer, either, any more than is a ball rolling down a hill, but I'll accept that others may.

Jim Caudill
12-21-2008, 03:15 PM
"It's a long way to tip-an-orrey".... reminds me of a posting made by Forrest a few years ago. Unfornutately, I was unable to find it and post a link. It goes on quite a bit to make the quote above a punchline or should I say "punline".

Evan
12-21-2008, 03:32 PM
Winding a clock is not information input, it is merely energy input. Setting a clock is information input.

dp
12-21-2008, 03:46 PM
Winding a clock is not information input, it is merely energy input. Setting a clock is information input.

Not setting a clock is accepting the current information. A clock doesn't compute anything, though (That is done by the observer who mentally converts hand position, or state, to approximate time of day). Especially one that has never been wound, and yet it's right twice per day which is probably more often than if it were running :)

There is no practical difference between winding a clock and turning a knob on the antikythera. The escapement and spring are analogs to that knob - they cause a train of gears to turn. Add a spring and escapement to the antikythera and you have a clock that is wrong about most things most of the time without human intervention.

BTW, we escaped from Oroville yesterday and made it back to Seattle between snow storms. All east-west highways are now closed. That was close!

Evan
12-21-2008, 04:19 PM
There is no practical difference between winding a clock and turning a knob on the antikythera. The escapement and spring are analogs to that knob - they cause a train of gears to turn.

Not the same. The key difference between simply winding the clock and setting the clock is determinism. Turning the knob on the Antikythera is accepted as input that determines the output via mechanical algorithms. Further, supplying the Antikythera with an escapement isn't the same as turning the winding stem on a watch. The Antikythera isn't a clock and continually turning the knob via an escapement in the reverse direction would cause it to display the same information as the date changed, in effect continually updating the calculation.

Turning the knob IS related to a source of information, namely the present date. A clock with a moon phase and day of week and month dial is much more equivalent to the Antikythera although much simpler. Add in the positions of the 5 known planets with the phase of the moon and corrections for the orbital anomaly and include future solar eclipses plus future time, date, and place of the olympics and you have the Antikythera.

dp
12-23-2008, 12:25 AM
This site has a section of movies - they're worth seeing. It takes some time but the machines are quite spectacular:

http://www.patek.com/