The Antikythera Mechanism Episode 7 – Making The Saros & Exeligmos Train

G’day Chris here, and welcome back to Clickspring In this video I make the gearing and support structure required for the eclipse prediction train of the mechanism Despite the fact that a solar eclipse is a reasonably common event, they do tend to be fairly localised So most of us rarely get to see more than one or two over a lifetime The result is that even though we have a complete understanding of what’s going on, the passage of totality over the surface of the earth retains its power to completely captivate our attention It is to say the least, a profound experience And it was perhaps even more so in the Ancient World Across many cultures, Eclipses, both lunar and solar were believed to be omens of great significance Foretelling the imminent death of a ruler, the outcome of a military campaign Or on occasion just providing a good excuse to justify a particular course of action Little wonder then that the Ancients chose to include an eclipse prediction function when constructing their all encompassing model of the Cosmos At the heart of the eclipse prediction method used, is what we call The Saros Cycle Yet another fortunate cosmic coincidence, where 223 lunar months after a given eclipse, a nearly identical eclipse will occur This leads to what’s known as a Saros series, where over a period of well over a thousand years, roughly 70 to 80 eclipses, will follow one after the other in sequence, separated by the regular interval of 223 lunar months The observation of any single eclipse from a given Saros Series, is all that’s required to identify the date and the time of all subsequent eclipses within that series, and so makes possible accurate eclipse prediction Now the Saros dial carried the markings for a large proportion of the eclipse series that were active during the period of the mechanisms construction I’ll cover this in more detail when I engrave the dial in a later video But essentially all the user needed to do to predict an eclipse possibility, was to crank the mechanism forward until the Saros pointer reached one of the marked cells The inscription indicated whether the eclipse was solar or lunar, as well as the nominal time of day that it was to occur And the month of the eclipse was simultaneously indicted on the Metonic dial Now since a lunar eclipse can only happen at full moon and a solar eclipse at new moon, the day of the eclipse could be easily identified By simply noting the day that the required lunar phase was to occur, or by cranking the mechanism forward and using the front dial to observe the day of alignment So Ignoring sign convention, and viewed from the perspective of the pointers, the Saros gearing calculation can be represented as follows Again, you’ll notice that the 53 tooth gear is cancelled out, and with simplification, the expression can be reduced to the following Which presents the behaviour of this part of the mechanism in a very intuitive way Because, you’ll recall from the previous episode that the Metonic cycle is 235 lunar months in 19 years So 19/235 is just another way of saying “one lunar month” We need 223 of these months to represent the Saros cycle, and we’d like them to be divided over a 4 turn dial A slightly less intuitive way of presenting it is as the number of days But this does highlight one of the defining characteristics of the Saros cycle: And that is that its not a whole integer number of days That 1/3rd of a day remainder, has an effect in particular on the location of the repeating solar eclipses The earth of course continues to rotate for that additional 8 hours, and so the location of the repeat eclipse is shifted about 1/3rd of the way around the Earth After 3 appearances, its back very close to the original location Which brings us to the Exeligmos pointer Its role is to indicate this 8 hour time shift information, advising the user to add either 0, 8 or 16 hours to the nominal eclipse time And again the gearing calculation can be simplified to something that shows exactly what’s going on We see 223 lunar months, the Saros period, multiplied by 3 Or as its otherwise known, The Triple Saros So where exactly did this deep astronomical knowledge come from? Well essentially from direct observation of the night sky It turns out that the Saros cycle, and in fact much of the other astronomical knowledge underpinning the mechanism was known to the Babylonians for a very long time prior to the mechanisms construction Clay tablets from the period suggest a culture that were serious watchers of the night sky With continuous detailed observations, that spanned generations

These observations were used to develop sophisticated predictive models, and much of the research supports the idea that at least some of this knowledge made its way into Ancient Greek culture So the astronomical heritage of the mechanism is relatively well understood But we know much less about its engineering heritage, or the workshop tradition that enabled its creation Currently, the best source of information is the device itself Because its features directly imply the existence of certain tool technology Now the exact nature of that technology is the big question, and we may never know for sure But we can certainly try a few things out and see what might have been the case So with that in mind, its time to open up a hole position that I marked back in episode 2 This is the pivot location of the E assembly, and its a perfect opportunity to make the very first hole in the mechanism using some of that ancient tool tech: The Pump Drill I’ll be adding a considerable amount of gearing under the E assembly in a later episode, that essentially supports it But in the meantime I need a temporary support and this bearing will do the job For this part of the train there are 5 wheel assemblies to make, as well as their supporting structures Starting with the E assembly, the scans show that the ring gear was fastened in place using rectangular pierced lugs and styled cotter pins Most of the mechanism uses simple tapered pins fastening, so this is different, and certainly gives this assembly a distinctive character Its important when installing the lugs that the wheel alignment be maintained throughout the process So to help with that, I used the small jig that served a similar purpose when making the main solar drive wheel B1 With a simple bushing installed, it can now serve in a similar role as a drilling platform for opening up the holes for the lugs, and also as an alignment tool during the fitting process That first hole serves as a convenient drilling guide for the hole that’ll receive the lug I turned the lugs to be a light interference fit in the holes, so a gentle tap with a hammer was enough to seat them tightly home in preparation for peening Ok, so with the first lug in place, the next step was to open up the matching rectangular hole,

so that the wheel can slide into position over the lug The alignment jig now becomes essential, making it possible to progress through the fitting process for all 4 lugs, whilst being quite sure that the wheel alignment is being accurately maintained throughout The cotter pins were then brought to a close fit with the lugs and pushed firmly into place OK, so with the E assembly complete for now, I moved on to the F assembly Which unlike most of the wheel assemblies made so far, rotates on a post support structure The wheel assembly consists of 2 wheels that are permanently staked onto a square hub Requiring the formation of the hub, as well as the opening up of the wheels to match the square Off camera I turned up the post and riser that make up the supporting structure for this assembly Its a simple disc spacer, and a post that passes through the main plate to be fastened with a pin and washer on the other side Next up is the bearing for the Saros pointer, which is yet another composite structure consisting of a rectangular block body and a separate spacer flange The Saros arbor was formed with a very light taper to ensure that the wheels were a firm,

well registered fit And the pivot sections turned to be a close running fit in the bearing Now Its helpful to consider the H and I assemblies together, since they share this most unusual pivot block as a common support structure The arbor for the Exeligmos assembly was constructed in the usual way, but I made a temporary and quite short version of the H arbor to allow for a specific depthing technique that I’ll show in a moment What I call the HI Pivot block is an unusual little idea, that will eventually require a small clearance notch to be removed in a later episode In fact it’ll remain loose in the mechanism until I can cut that notch For now though, its enough to simply depth and plant the 2 assemblies in a suitable place on its surface, and form the basic profile The main plate was then opened up, and the square bearing hole formed These two positions are essentially absolute locations, governed by the dial artwork And depthing could have been conducted from these positions before filing out this square hole But at the risk of the position almost certainly moving as the square was formed, and so compromising the depth So its worth pointing out that the Makers choice of putting the F assembly on a post, rather than an arbor, permits the use of a very simple and practical depthing tool This means that depthing can be left until after the square bearing has been fitted, and so any position error can be accommodated without issue The temporary version of the H arbor, means the HI pivot block can be used in a similar manner A small pin ensures that it pivots from the Exeligmos position, and the block can then sweep over the surface of the main plate, to find the correct depth of engagement for the H assembly Again removing the risk of the Saros pointer bearing being poorly positioned when filing the square Now of course the very compact nature of the machine

means that vertical clearances also need to be accurately set I mentioned in Episode 3 that I think the composite structure of the assemblies, and in particular the use of spacer components, was the secret to achieving the incredibly close clearances implied in the wreckage And here’s a perfect example of what I mean The height of the saros pointer assembly is set by the height of the bearing in which it sits, and that height can be easily set by this little spacer I originally made it slightly oversized, but its a straight forward job to set a clearance of just a few hundredths of a millimeter, by simply abrading the spacer until the barest daylight remains between the adjacent assemblies No special tools, or absolute measurement are required Just keen eyesight, and a flat abrasive surface The entire mechanism could have been set to very close clearances using this simple trial and fit method, and I’ll continue to use it as I move forward with the build Ok, so with the depthing and clearances set, the temporary H arbor was removed, and replaced with the permanent full length version This arbor threads up from underneath the plate, and so requires that a fastening pin be positioned just above the wheel and pinion And I used the bit component of the pump drill on its to form the hole It is much slower than using the driving mechanism, but using it by hand gives greater control over the hole position The other assemblies were pinned using the same technique, and with that complete, its time to put it all together and see how it performs In the next video I’ll make a start on one of the more remarkable sections of the mechanism: The gearing that models the lunar orbit Thanks for watching, I’ll see you later