Failures, disasters, fractures and lithic technology

ALLISON BYRNE: Good evening everyone. I’m Allison and I’m secretary for the Canberra Archaeological Society, who with the Friends is putting this lecture on tonight. I do hope you enjoy it. Oliver came into my life when I, as a mature student started to do archaeology, way back in 2003. He was one of my first tutors and we were just reminiscing where we went with that. It was a lovely time. I think it was actually the best years of my life, besides having six kids. It was really great.

Anyway, I’ll just give you – you’ve probably read it anyway, ‘the lessons learned from shipwrecks, collapsed buildings and other disasters are surprisingly useful for understanding how prehistoric artisans produced flake implements. Dr Oliver Macgregor explains how the principles of material fracture can be applied to the manufacture of stone tools’. Without further ado, let’s welcome Oliver. Thank you very much.

OLIVER MACGREGOR: Thank you and good evening. I’m going to talk to you this evening about the endeavour to understand flaked stone tools in the archaeological record, and to focus on how that endeavour has to be and is a multidisciplinary task where we as archaeologists draw lessons from our colleagues in other fields such as physics and engineering and material science, who are asking similar related questions to the ones we are.

Stone artefacts are obviously massively important in the story of humanity. For most of the time that we’ve been on the earth, the hardest material, the most durable material that we could access was stone; and the sharpest cutting edges that we were able to produce were the cutting edges that we could make through breaking apart pieces of stone. Stone artefacts are consequently very important for humanity, and they’re consequently a great focus for archaeologists in understanding the human story.

Every human society that uses stone artefacts faces a number of challenges. Their entire technology revolves around breaking stones into smaller pieces to produce sharp cutting edges, which means that at every point in the process, you’re faced with the twin challenges of obtaining stone and then continually re-obtaining and re-procuring stone as you progressively exhaust the stone that you have to hand. Obtaining stone is a cost because not all stone is particularly well-suited to producing flake stone artefacts. Most stone is very poor for the purpose and it’s usually just the finer grain, glassier, more brittle stones that are of value. Those are distributed around the landscape in an uneven fashion, often in an unpredictable fashion, and so there’s a cost of procurement associated there.

There’s a cost of maintaining the tools that you have in your toolkit because as you use these sharp cutting edges, they degrade and blunt surprisingly rapidly, and stone artefacts break as you use them so you have this ongoing attrition of the tools that are being used. Therefore, it is a necessity to continually procure new stone and to convert that into new cutting edges. Stone artefact technology therefore is an ongoing process. It’s a struggle to continually procure and consume resource that is always being perpetually exhausted, so to speak.

Just a couple of illustrations of the process of what it is to make stone artefacts. On one level, it’s as simple as smashing rocks apart by the application of force. Any stone artefact no matter how complicated or simple it is results from a number of these events, a number of events where force has been applied to a piece of stone and pieces of rock have been flaked off that parent piece of rock. Even very complicated artefacts like a point – this is a tula adze from Central Australia [shows slide] – even these very complicated artefacts, they just result from multiple, multiple flaking events which produce all these little scars and marks and gradually shape the artefact into that form.

This is the process in its simplest terms. It involves applying force to a point on a piece of stone. Just some terminology: the surface that you’re applying force to is the platform surface, the piece of rock that you are breaking apart we refer to as the core, and the pieces that are broken off that core we refer to as flakes. It’s usually done through a percussive force by applying force with a hammer, which is usually another piece of rock, to that platform surface and applying force until that loading overcomes the strength; overcomes the fracture toughness of that piece of rock. A fracture initiates at the point of contact there, it travels down through the piece of stone and when it terminates at another surface – so we’re looking at this process in cross section here – a small piece is flaked off. The piece that comes off is the flake, the piece from which it was detached is known as the core.

Because this is a fracture process, it means that the task for us as archaeologists is to try and understand the nature of that fracture process. Prehistoric humans understood it very well because their lives and their livelihoods depended upon it. They had to be able to flake stone in a reliable, predictable, efficient manner. In certain circumstances, there was an incentive for them to do that as predictably and efficiently as possible, particularly in situations where the cost of procuring a stone were high, both in terms of the time you would have to invest, the energy you would have to invest in transporting stone across the landscape, and consequently, the time and energy that that would take away from all the other tasks that you had to do in order to survive.

Every moment that you need to spend going out and finding more rock to replace the stuff that you’ve exhausted is time that you cannot spend procuring other resources such as food and water, and all the other things you need in order to survive. Prehistoric humans understood this process in a very sophisticated way, and it therefore falls on us as archaeologists that we have a necessity to also understand that fracture process in a sophisticated way as well, so that we can understand the behaviours through which they produced their artefacts. We can work our way backwards from the archaeological record, observing the artefacts that we find, recognising the behaviours and the strategies that prehistoric people were employing, and then making statements about the situation on pressures that those people were under in order to employ those strategies and those behaviours.

This fracture process is something that’s not particularly familiar to any of us in the modern world. None of our technologies really operates in terms of fracture. None of our technologies operates in terms of reducing material in this way. Most of our technologies involve constructing things, building up, rather than reducing down. It requires us to operate in a fairly unusual and unfamiliar landscape. We can learn from other disciplines who also need to understand the fracture process, in particular engineers, physicists and material scientists; the people who we rely upon to construct things that we use every day. They need to understand fracture because they need to be able to be sure that these structures that we use are not going to break apart and fall down. We need to be confident that when we build a structure such as this Museum, it’s not going to suddenly crumble as we sit here.

The way in which we can utilise those fields in order to understand the fracture process is a massively huge and complicated one. For the purpose of this talk, I’m going to boil it down into one very small case study, and that is understanding a peculiar human behaviour known as ‘overhang removal’. Overhang removal is a strange, little preparatory strategy that people sometimes used prior to striking a stone flake. It involves removing tiny, little flakes from the edge of the core, so flakes that are very small in proportion to the flake that is about to be struck. Taking those off the edge of the core first and after that overhang removal step has been carried out, most flake scars have been created, then applying force further back to produce the main flake.

This is a strategy that can be fairly easily recognised in the archaeological record because the flakes that are produced have these overhang removal. They have the overhang removal scars still present on their dorsal surfaces. [shows slide] This little scar here corresponds to all these little scars illustrated on the dorsal surface of that flake. It’s something that shows up very readily in the archaeological record. We can easily recognise when it has and hasn’t been applied or used as a strategy.

It doesn’t always get used. You find it on some sites but not others. You find it on some materials but not others. It’s a big puzzle as to why people sometimes choose to do this. They sometimes choose to employ this preparatory step prior to striking a flake and sometimes they choose not to do this.

Why does it happen? In order to start to answer that question, I’m going to tell you a story about something which is not related seemingly at all. This is a story about an aeroplane that was flying from Colombia to New York in 1990. It was Avianca Flight 52. It took off from Medellín. It was travelling to New York but it never quite got to the airport. The weather in New York that day was very poor. There were a lot of planes lining up to land at the airport. This flight was placed in a long-holding queue. When its turn finally came to land, the visibility was poor and the plane couldn’t see the runway as it was coming into land. It had to suddenly turn on its engines again, accelerate upwards and execute what is known as a ‘go-around’, so turning around in a circle to have another go at landing.

Ordinarily that’s not a problem for an aeroplane, but for Avianca 52, it was a disaster because they were at this point virtually out of fuel. The reason that they were out of fuel is a long, complicated and rather terrifying story. It’s not entirely relevant to the talk so I’m not going to tell the story to you now, but I think what I’ll do is after the talk is finished, as you’re preparing your questions, I’ll tell you the story then because it’s not archaeological but it’s very interesting nonetheless.

The result of that is that as they were executing the go-around, they ran out of fuel. They glided to the ground and they crashed in a very affluent area known as Cove Neck on Long Island. There was an emergency call which is blackly humourous, which the call to the police was a lady who said, ‘I live in Cove Neck in Oyster Bay and there is a plane crash in our yard, in front of our house,’ in a very matter of fact voice as though there’s nothing particularly unusual about the fact that a 707 might be able to land in your front garden.

The crash of Avianca is an interesting story, but the aspect that I want to focus on, which is of relevance to the talk, is what happened after the crash. When the investigators got to the site of this crash, they found that there had been no fire in the aeroplane because there was no fuel, so there’s no explosion. That was a good thing. Survivability you would think would be fairly good in a crash like this, but they discovered that one of the curiosities of this crash is that inside the cabin, a lot of the seats had detached and broken away from the floor. I won’t go into the details, but that’s obviously a very bad thing to happen during a crash, for the seats to suddenly come free and be able to move around inside the cabin. It was a bit of a disaster in terms of the human lives.

The question then was, well, why did that happen? Seats at that time were designed in a way that was thought to be sufficient to withstand the force of a survivable crash. In order to do that, the engineers calculated the acceleration or the deceleration that you would experience during a crash that was not severe enough that it would kill you through deceleration alone. They tested these seats to nine times the acceleration of gravity, and they tested them by applying a load to the seat frame, so a full size proper seat; a full size experiment that would apply a load to the seat frame equivalent to the force that the weight of the seat and a human sitting in it would generate under that 9g deceleration. All seats in planes at that time were designed to withstand that test, so they were over-engineered to withstand that 9g deceleration. They were stronger than they needed to be under those circumstances.

We have then a question, a puzzle of why did the seats in that crash – which should have been well under 9gs – why did those seats break off and why did so many people become injured as a result? Leaving Avianca 52 for a second, to return to the world of stone artefacts, our other great imponderable why, why does overhang removal occur?

If we think about the flake production process in a little bit more detail, the stresses and the forces involved during ‘fracture initiation’ are on the face of it fairly straightforward. The hammer is essentially a spherical piece of rock [shows slide]. It contacts the platform surface in a circular contact area because it’s a flat surface contacting a spherical surface, so the contact between them is flat as the hammer deforms. The force is almost always applied at an oblique angle by the person who produces the flake. It’s almost always applied at an oblique angle across the platform like this [shows slide]. That means that that angle therefore produces two components of force that act into the rock.

First of all, there’s a downward component, a direct component of force that acts straight down into the platform. The other component of force is a tangential component that acts across the platform. Essentially you’ve got most of the force of the hammer pressing down into the platform. Some of the force of the hammer ripping across the platform and pulling the platform towards the edge of the core, so towards that free surface of the edge there [shows slide].

Is that clear to everyone? Good. Excellent. If at any point I lose you, just throw up a hand and shout out.

The direct component of force is no problem at all for a piece of stone. [shows slide] That component of force is acting straight down. All it does is compress the stone. It just produces a zone of compressive stress under the point of contact. Stone is extremely good at withstanding compression, which is why a Gothic cathedral’s roof can stand up on tiny, thin, little spider webby-type pillars because those pillars, though they’re very thin, they can withstand a huge amount of compression that entire roof is bearing down on them.

The other component of force though, the tangential component that’s pulling across the platform, that has the effect of producing a zone of tensile stress directly behind the hammer. Because it’s tearing across the platform, it’s stretching out the platform surface so the intermolecular bonds of the material at that part of the rock are being stretched apart. Tension is a big problem for stone. Stone is very weak in tension, which is why the roof of the Gothic cathedral has to be arched. You can’t have a flat ceiling or anything like that. That’s why cathedrals are narrow relative to their height because they can’t have much tension operating in a stone structure.

The tensile zone there is where the material fails [shows slide]. This is why production of stone artefacts is possible, because setting up that zone of tensile stress enables a knapper with not much force at all to break apart a piece of rock which on the face of it would seem to be quite a strong material.

If that’s the case, if all it takes is a zone of tensile stress being set up by the hammer, then why should overhang removal make any difference to that process? If what you’re doing is setting up a zone of tensile stress there [shows slide], and that starts the fracture going, and if you think about it, as that fracture advances, that force which is still acting up there is still producing a zone of tensile stress at the fracture tip. It keeps that fracture propagating through setting up that zone of tensile stress as the top of the flake is ripped away from the rest of the core. That stress should be more or less independent. It should be independent to what’s going on out here.

If you think about breaking apart a block of chocolate, which is essentially the same process, you’re bending the centre of that block of chocolate and setting up a tensile stress along the top of the block. [shows slide] It doesn’t really matter what’s happening over at the end of the block of chocolate there. It doesn’t really matter if the end of the block of chocolate is there or if the block of chocolate ends back here or over there. Breaking it apart in the centre is just as easy either way.

Why would it matter mechanically if we’ve got a piece of material that’s been stripped away from the edge of the core prior to removing the flake? It doesn’t seem to have any straightforward mechanical advantage at all. For this reason, archaeological explanations of overhang removal have largely focused on non-mechanical mechanisms for why it might be beneficial. The explanations usually involve overhang removal functioning as a way for knappers to increase their precision and their accuracy of hitting the platform surface.

In brief, for a number of reasons, it’s problematic if you hit a core too close to the edge because you’re usually hitting it then with too much force and that edge just shatters off in a rather chaotic way. There’s a benefit for a knapper not to hit the core out here at this little, narrow edgy bit [shows slide]. The usual explanation for overhang removal is, well, it’s a way of just removing that bit by going tap, tap, tap, tap, stripping away that material that would be problematic if you inadvertently hit it, and therefore acting as a failsafe to make sure you don’t hit it when you apply a much larger force to take off the primary flake, which is quite possibly true.

It’s a little bit problematic in my view because prehistoric knappers were often immensely skilled and they were able to produce artefacts that through replicative experiments today, we know that it takes many years of training for one of us to develop to a comparable level of skill as some of these prehistoric knappers. We know this because people do this as a hobby. There’s lots of people who have devoted huge amounts of time in order to become expert knappers themselves. It takes a lot of skill to replicate the sorts of artefacts that some prehistoric technologies were able to produce. Once you get to that level of skill, you’re operating in a Tiger Woods level of being able to have hand-eye coordination.

The explanation that you would have overhang removal as a way of increasing precision is a little bit problematic in my view; in those circumstances that you essentially would not need that training wheel’s mechanism once you’re at that level of skill, but knappers who are producing very sophisticated artefacts in the archaeological record are also employing overhang removal. There’s a bit of a mismatch problem there. They seem to be doing something that if it were just about skill, they’re at a level of skill that they would not need to employ that aid to accuracy and precision.

That leaves us with an issue of trying to explain it in mechanical terms and encountering this problem of, well, it doesn’t seem to have any effect on this tensile stress field, so why is it having any effect on whether or not the force of the hammer produces a fracture and therefore successfully removes the flake? The problem there is, well, maybe we are looking at fracture in a slightly simplistic manner. Maybe we’re not thinking about the complexities of it. Maybe there’s something else going on here. It turns out that the engineers who designed aeroplane seats had fallen prey to that same problem as well.

Avianca 52 was not the first airline crash where seat failure had been observed. Earlier in 1983, I think it was 1982. Sorry. I want to get the details right because if you get the details wrong of something like this, you can be sure someone in the audience is going to be an airline history buff and they’re going to instantly correct you and not believe a word you say afterwards. In 1982, there was an aeroplane called Air Florida [flight] 90, which took off from Washington, DC on a very, very snowy, cold day. It took off. It didn’t have enough power as it took off and it crashed almost immediately. Basically, it failed to gain any altitude at all. It suddenly stalled out, crashed down into the Potomac River, hitting a bridge along the way.

Similarly with that – this was an earlier crash than Avianca, which was in 1990 as you recall – in that crash, it was observed that seats had broken off on impact. It probably didn’t make any difference to the people who were on that plane in this case, but nevertheless, people observed this and they set up teams of engineers to investigate what was going on. In order to do that, the engineers set up as realistic an experiment as they could. They built a little replica, a full-scale replica of a section of the aircraft cabin and they accelerated that up to a speed along a little track and then slammed it into a barrier so that it would decelerate rapidly. Again, mimicking the deceleration that would occur in a survivable airline crash. They found that the forces that were involved were not straightforward. The deceleration didn’t happen in a regular manner, operating on the entire seat frame at once as had been previously expected.

What they found when they did this experiment and they filmed it with high-speed cameras so that they could watch the process in slow motion is that what decelerates first is the cabin floor because that’s the bit that’s actually attached to the front of the plane, and that’s the front of the plane that’s probably what’s impacting first. The force of that impact has to travel back through the cabin floor. What happens is it travels in the form of a shockwave, a transverse wave where the floor bubbles up, basically, like a piece of string that’s being plucked, and the shockwave travels backwards down the length of the cabin.

When it encounters a seat, the very strange thing occurs, which is that the inertia of the seat itself becomes a problem for the structural integrity of the seat. The seat has mass. The seat and the person sitting on it, they have a combined mass. That mass resists this upward force [shows slide]. It resists the upward movement of the floor. That inertia holds the chair stationary, the main part of the seat stationary, so all that deflection has to be absorbed by the seat post. As this bit is stationary, the floor is rising up. The seat post has to take all that deflection. It gets deflected beyond its limits. It’s therefore experiencing stresses that it’s not designed for. That’s where the failures occur. When it reaches the back seat, the same thing happens. They found it fairly predictably, these seats failed and they failed in the same way that had been observed on Air Florida 90.

The process of seat fracturing is not a static process. It’s not a straightforward process. It’s what engineers call a ‘dynamic process’. A dynamic process is one in which the inertia of elements within that system become of prime importance. No longer is the situation purely about the strength of the materials involved. The situation becomes much more about the inertia of pieces, their ability to move and respond to forces because those forces are being applied rapidly, and they’re not being applied slowly and gradually. We’ve got rapid forces; therefore, the ability of things to deflect and their inertia becomes of prime importance.

The other key point here is that that was only understood after a replicative experiment was carried out in this way. A dynamic system is very difficult to predict, so the dynamic nature of this process would have been very difficult or impossible to predict on paper. It was only after we had this experimental verification that the nature of the process becomes identified and therefore understood.

Could flake production be a dynamic process in the same way? It makes intuitive sense that it might be because it has all the hallmarks of a dynamic system. It involves loading one piece with another piece of rock, and it involves that loading creating stress that leads to a fracture. The loading is applied very rapidly. We’ve got a percussive force where the loading is hitting in the space of a few milliseconds, the hammer is contacting the core, the force of that contact is deforming the hammer as it sits on the core surface, and then after a couple of milliseconds, that force will be gone. If no fracture occurs, the hammer will bounce away and so we’ve only got a small amount of time for a fracture to initiate if a flake is to be successfully removed. We have rapid loading and therefore, we could expect that the situation of fracture would be a dynamic one in the same way that the fracture of airline seats was dynamic.

The key issue here is that for a fracture to start propagating, what needs to happen is that fracture needs to open up a little bit. By definition fracture is where the material started to break apart. We need to get those – if you think about it at a molecular level – those little molecules have to stretch apart and then they have to break. The first intermolecular bond on the surface has to break and then the next one down has to break and then the next one down has to break. By the time you’ve got down through a few million molecules, the ones on the top have to have opened up quite a long way in order for that stress to propagate down to the fracture tip, and for the next intermolecular bond to break.

At a macro scale, it looks a slight exaggeration like this [shows slide]. In order for the stress to be maintained at that crack tip, we need to continually have the top of the fracture opening. The fracture has to be gaping open in order to start and in order to continue propagating. For that to happen, it means that all this material between the fracture and the free surface of the core has to deflect outwards. We’ve got an area of material that needs to deflect, it needs to deflect rapidly, and it’s an area of material that has mass and therefore has inertia. It’s going to resist that deflection if it needs to deflect rapidly. We have the hallmarks of a dynamic system there where inertia of that piece of material might be quite important.

Theoretically, overhang removal may function to decrease the inertia of that piece of material because we’re taking material off that free surface. We’re stripping material back from that surface. We’re reducing the volume of that material, therefore reducing its mass, therefore reducing its inertia. In theory, you think, well, it could have a mechanical advantage therefore in that it allows that platform surface to deflect a little bit more easily, and therefore allows the fracture to propagate a little bit more easily.

That’s all very well for theory of course, but does it actually have enough of an effect in reality for this to be a valid explanation as to why prehistoric knappers employed this as a strategy? As with airline seats, the only way we can know this is through an experiment. What I went away and did is I set up an experiment where the objective was to control as many factors as possible. Control the amount of force that’s being applied, control the shape of the core that is being flaked, so that a lot of replicates could be done of cores, with and without overhang removal.

The cheapest way to do this – let’s be honest, science is often driven by money particularly when you’re a graduate student, so money was at the forefront of my mind – the cheapest way to do this was to use glass rather than stone. Glass is a brittle solid which essentially behaves the same way as stone. A lot of experimentation of stone artefact research is done on glass. Glass has the advantage that you can machine it fairly easily. You can cut it into shapes and control those shapes fairly readily. You can make as many replicates of the different shapes as you want.

To control all the other variables, I held these cores in a vice so the resistive force overall is controlled each time. The hammer used was a spherical steel ball that was dropped from an electromagnet, which means that the force is exactly the same every time; the angle of impact is exactly the same every time. You can reload this experiment time and time again with each type of core, and you can vary where the force is hitting the core each time. [shows slide] You hit it there then with the next core, you reload a fresh new core, hit it there, hit it there, hit it there, each time with a fresh new core so that each time you’re looking at how the fractures behave, hitting the platform surface at different points, and you’re just looking at how the fractures behave relative to the shape of the core and the point of force application.

I’m going to talk today about just two experimental series that I did, both of which have an exterior platform angle of 70 degrees. That’s the angle between the platform surface and the pre-surface of the core. Seventy degrees is a fairly typical angle that you find on cores in archaeological sites, so it’s a reasonable replica of the typical situation that a knapper would encounter when they’re striking flakes off a great, big core. The two series are: one has a flat free surface over here. That’s mimicking a core where the overhang has not been removed. The other one had a little concavity, five millimetres wide, 20 millimetres long, mimicking the small flake scars that you would have if overhang removal had been carried out on that core. They were loaded into the apparatus. We replicated it many times to vary the platform thickness, at which the force was applied each time, and then we can compare the two.

Looking at the core without overhang removal first [shows slide], this graph shows a scatter plot of the length of the flake that was produced on the Y axis versus the platform thickness at which force was applied on the X axis. I just put a little diagram of the core down here, showing the points at which force was applied in this experimental series. Bear in mind each time, you’re reloading a fresh core. It’s not that these forces are taking multiple flakes off the same core. Each time that one has a fresh core with that exterior platform angle, and this hit back there has exactly the same exterior platform angle and size of core.

This result was fairly unsurprising and mimicked previous experimentation that had been done in America. What you find is a fairly straightforward relationship that as you increase the platform thickness, the length of the flake increases. That’s a fairly straightforward and easy to understand effect because the further back from the platform edge you place the force, the further the fracture propagates through the material before it terminates out on that free surface. The further back you put the force, the bigger the flake you can produce.

It increases until you reach this threshold here [shows slide], where if you place the force any further back on the platform, absolutely nothing happens. Once you get to that point, no flake at all is produced. You get a failure of that impact to produce a flake. All it does, is it produces a little fracture feature in the surface called a Hertzian cone, which is a little conical shape fracture, just half a millimetre or a millimetre long. If you’ve ever fired a BB gun at a window, you’ll observe this cone fracture. It’s a little cone that propagates through the material. That’s all that happens at that point. You just get a little cone forming there and absolutely nothing else happens. The hammer just bounces off. No flake producing fracture occurs.

For a knapper, it’s a double failure because not only have you not produced a flake but you’ve created that little cone fracture in the material. Murphy’s law has it that when you’re knapping a piece of stone, some point further down the line, that little cone fracture will come back to haunt you. You’ll be flaking it from a different angle and the fracture you produced will propagate into that cone fracture and deviate off in all sorts of weird and wonderful ways. It’s not a good thing when failures like that occur. You get this region of viable platform area basically where that’s your viable platform, and above that, it’s not a viable platform anymore.

Just reminding us what the two series look like. We’re now looking at the second series with the overhang removal scars machined into their surfaces. This is what the situation looks like there [shows slide]. I’ve left the first series on as blue dots and the second series that we’re now looking at is red dots. Down here is a scale image of the core. That amount of material has been moved by the removal of that overhang removal scar, so the little concavity that mimics the overhang removal scar has removed that much material. We can only hit the platform above that point.

First of all, don’t worry about those two points down there. They’re very interesting but they’re not particularly relevant to this talk. Essentially what’s happening at that point is the shape of the concavity that was machined into the core has an effect on the fracture as it propagates. If you hit the core too close to that platform edge, the square shape of the bottom of that concavity produces its own little stress field. When the fracture propagates past that, it suddenly deviates off and ‘hinge terminates’ or ‘step terminates’, so it dives out towards the bottom of that concavity.

That incidentally was what these experiments were originally designed to study before I started getting interested in overhang removal. You can read all about that in a previous publication that I can email you the details of if you’re interested. They’re not particularly relevant to this talk. What we’re interested in is the other end of the spectrum, which is what happens at the top, the maximum values of platform thickness.

Interestingly, we find that overhang removal does increase the maximum platform thickness at which you can apply force and successfully remove a flake. Without overhang removal [shows slide], we could only go up to here before we start experiencing failure. With overhang removal, we can go up to here before we start experiencing failure. On this scale of core, we’ve got a few more millimetres of viable platform. That was a nice result. It confirmed the hypothesis. Overhang removal is making it easier for that force to successfully remove a flake.

Perhaps even more interestingly, the length of the flake that you can remove is greatly increased as well. That small increase in platform thickness, the small increase in the viable platform area from which you can strike a flake results in quite a large increase in the maximum length of the flake that you can get off. In this first series, as you start to get up to that critical threshold of platform thickness, the flake starts to tail off in their length. Presumably, it’s starting to run out of energy a little bit once you get to that critical point. That was the maximum length, just slightly over 18 millimetres long in this case.

The same effect happens with overhang removal but at a far higher point. You get an extra 20-something millimetres of flake length. You’ve increased the maximum flake length by about 30 per cent, so quite a substantial gain in the maximum potential flake length that you can strike, relative to the amount of material that you’ve taken off through the overhang removal, and relative to the amount of extra platform thickness that you can utilise.

Is that all clear to everyone by the way, those two graphs? Fantastic. Good. You’re much smarter than most undergraduates that I taught. That was just a joke. If any undergraduates are listening on this recording, they all understand it as well. I’m just working the demographic of the room there [laughter].

The points to take away from that experiment are that overhang removal now doesn’t need to just be viewed as an ad hoc cultural behaviour. It’s not a strange cultural trait or it’s not a stylistic trait. It’s not just a mechanism for increasing precision, although that still could also be the case. This is not in opposition to those early hypotheses, but it doesn’t just have to rely on those explanations, which are very difficult to test in any case of course as you can understand. It means that we now have a fairly straightforward mechanical explanation for overhang removal as well that overhang removal creates a fairly straightforward mechanical advantage for a knapper. It increases the knapper’s options, essentially.

It means that by carrying out this fairly straightforward little preparatory step of tapping off the overhang removal flakes, which is a very easy process and entails no risk for the knapper, there’s no way you can fail to do overhang removal, basically. It’s a fairly straightforward preparatory step. In doing that, you increase your options in terms of the types of flakes that you can produce. You can produce a much larger flake without, and this is the critical point, without increasing the amount of force that you have to apply. You can hit further back on the platform and produce a bigger flake without having to increase how hard you’re hitting it.

That’s important because as a knapper hits a piece of rock, there’s a limit to how much force they can exert upon that rock for a number of reasons. There’s the limitations of your strength and the weight of the hammer. More importantly, the limitations of how securely you can hold that core. If you’re holding it in your hand, there’s a limit to how much inertia you can give to the core essentially as you’re striking it. If you hit it too hard, your hand is unable to support the core so it just bounces away from you. There’s a limit to how much force you can input into that system.

Knappers can’t necessarily increase the amount of force that they put in ad infinitum, and of course that problem gets more and more pronounced as the flaking process continues. If you start out with a very big piece of rock, it’s fairly straightforward because it’s got a huge amount of inertia on its own so you don’t have to hold it very still. Once it gets very, very small, smaller and smaller, it has less inertia. Its ability to resist the hammer is less, and so its tendency to bounce away rather than being held still as your hitting, decreases and decreases. As you continue core reduction, that problem of limitation of the amount of force you can apply becomes more and more and more pronounced.

This means you can sidestep that a bit. You can employ this little hack, if you like, to get around that problem to an extent. By taking off a bit of overhang, you essentially negate the need for that extra amount of force to be applied, and you increase your options. Just with the force that you have available, you’re suddenly able to take off a larger flake, which is potentially what you needed to take off.

We have a mechanical explanation for overhang removal, and it means that it is a strategy that makes sense in terms of economising economical use of stone. It’s a strategy that enables a knapper to get around the problem of having to discard a core at a certain point rather than if you wanted to continue flaking and you’re suddenly running out of the amount of force that you can apply to that core, rather than having to discard it, being forced to discard it and go off and find some replacement stone at that point; overhang removal means you can prolong the reduction process a little bit more. You can remove the overhang so the force again that you’re able to apply is sufficient and you can continue flaking for a little bit longer.

It’s a strategy of prolonging the reduction process, and therefore it’s a strategy of economising your usage of stone, of delaying that point at which you have to discard and go out into the landscape and waste time and energy in procuring replacement stone material.

QUESTION: The amount of observation and intelligence to actually realise that is mindboggling.

OLIVER MACGREGOR: Yes, it is, isn’t it? The interesting thing is that when you teach someone how to knap or when you learn how to knap yourself, you start to do this after a while – I won’t say instinctively but it seems that just through an iterative learning process – you start to employ this as a strategy.

QUESTION: Would that be a positive to think about?

OLIVER MACGREGOR: Yes. That’s been my limited experience of teaching students how to knap. Other people in the room may have more experience of this, but it seems that, yes, you do after a while, you reach a point where people just start doing this without having being told to do it, basically. There is an iterative feedback process that, as you’re flaking, you learn what works and what doesn’t. I don’t know if it’s occurring in that way prehistorically or if there’s more conscious planning involved. It’s a very difficult question to be able to consider or to have the potential to answer.

People spent a lot of time flaking stones. People were able to experiment quite a lot, I guess. If you’re brought up in a tradition where you have to flake stone routinely, you’re doing it every day of your life, and all these other things in your life depend on your ability to be able to do it successfully; it’s plausible that this behaviour could occur just through an iterative learning process rather than through a particular deep understanding, at a material science level of what’s going on.

As an epilogue, I know you’re all curious to hear the end of the story about the airline seats. You’ll be relieved to hear that after they did all those experiments, they then redesigned airline seats and put them through exactly the same experimental procedure; so loading them on to a cabin floor, shooting it down the track and then decelerating it rapidly and filming the result. After trying a number of designs, they came up with this very simple idea which they call the 16g seat. It’s called the 16g seat because it’s now able to withstand a force of 16gs rather than 9gs. They’re planning for a much, much more severe crash because they think that the human body is actually able to potentially survive a 16g deceleration now.

What they did is they changed the material of the seat post and very cleverly, they altered the design of the rear seat post so that it’s now curved like that. What that means is that when the shockwave comes along, the front seat post is more able to compress as the shockwave passes through it. It’s a more malleable material now so it’s able to compress and deform as that shockwave passes underneath it. The back seat post just bends and increases its curvature and absorbs the deflection that way. After the shockwave has passed and the actual seat itself starts to move forward – so as the seat follows forward as the rest of the plane is decelerating – that seat post then extends and straightens out so it gradually absorbs the force, so it decelerates the seat much more gradually than it would otherwise decelerate.

It has those twin benefits, both of which are very beneficial in a crash. If you get into a plane nowadays and observe the seats, you’ll find that they’re all like this [laughter].

QUESTION: It won’t get off.

OLIVER MACGREGOR: That’s right. Exactly.

QUESTION: Why didn’t they do that back in 1982?

OLIVER MACGREGOR: Yes. Engineers have this rather grim term which they talk about, ‘design by body count’, which is that design changes only really get implemented once enough people have died. This is unfortunately a fairly good example of that rather nasty mechanism at work.

The first crash that happened after these seats were introduced was a crash in England at a town called Kegworth near Manchester. This is known as the Kegworth [air] disaster. If you’re an English person, you’ve probably heard of it. This happened in 1989. It was a British Midlands flight that crashed quite close to an airport actually. Its engines both failed. It was the same type of aircraft, the 737, as the other two crashes. Its engines failed. It glided down to the ground. It fell just short of the runway that it was aiming for. It crossed – I think it’s the M5 motorway so it just missed the M5. If it had landed on the M5, it would have been an absolute disaster of course. Unfortunately, on the other side of the M5 motorway, that’s quite a steep embankment that it ploughed into. It was a very, very nasty high-impact crash where it went into this steep embankment and stopped immediately.

So you see a much more violent crash than Avianca 52 where the whole thing has broken apart, but it had the 16g seats, and as a result, unbelievably more than half of the people in this aircraft walked away from it, sustaining only fairly minor injuries. The survivors were fairly well off. A hugely powerful crash, but as long as you weren’t in those critical bits that broke apart, it was very survivable.

It’s also an interesting crash because they were able to interview the survivors almost immediately. They had a team to get there to interview them the very next day while they were still being discharged from the hospital, to ask them what their behaviour was directly before and during the crash. As a result of that, they redesigned the brace position that is now advised in aircraft. The brace position now is different from when it was before because they found that the type of brace position that you adopt has a bit of an impact on the injuries you receive.

The other little footnote to this crash that I find interesting is that a lot of the passengers just flat out forgot to adopt the brace position during the crash. They all knew it was going to crash and they’ve been advised by the air stewards to brace, but when they interviewed them, a lot of them said, ‘Well, we just plain forgot,’ because the human brain is not very good under those situations of high stress.

Anyway, Kegworth was the first crash with the 16g seats. They functioned very well, and as a result of the Avianca crash, there was a huge political pressure to retrofit these seats into older aircraft as well. Kegworth was lucky because it was a brand-new aircraft. This was one of its first flights so it had the new seats. After this, they said, ‘We have to retrofit all aircraft.’ They were pretty sure that through the [19]90s that occurred, now virtually all aircraft have the new seats, unless there are a few hiding out in very, very benighted parts of the world. They haven’t found any yet. That’s the good news ending to that story.

The general message here is that archaeology in many respects is a multidisciplinary field. Stone artefacts I think are particularly a multidisciplinary field. I think we have a lot to learn from our colleagues over in engineering and material science. I think we’ve been fairly slow to do this, but in the last few decades, it’s accelerating at pace. I think this is very much the future of stone artefact research. It’s getting out and talking with our colleagues in other fields and absorbing the ideas that they have and letting them inform our research.

I’d like to acknowledge my two supervisors: the wonderful Peter Hiscock now at the University of Sydney; and Zbigniew Stachurski who’s in the engineering department here at the ANU [Australian National University]. There are too many other fellow students who I went through university with to acknowledge on these slides. I’ve just limited it to Chris Clarkson, who is far and away the most influential mentor that I had during this research.

That is the end of the talk. While you guys are mustering your questions, I have a bit of a question and answer session, but I promised you that I would inflict the story of Avianca 52 on you in all its horrible detail. As long as you don’t have any specific objections, I’ll spend the next ten minutes doing that.

Avianca 52 as you recall was flying from Colombia to New York. It had a very experienced crew. Its captain was a man named Laureano Caviedes. He was 51 years old. He had been flying for 16,000 hours, which is a huge amount of experience for a commercial airline pilot to have. His first officer was a man named Mauricio Klotz, who was 28. They had a flight engineer on board, who’s called Matias Moyano, who’s 45, who had been with the company for 20 years and also had a lot of flying hours under his belt. A very experienced crew. It makes what happens next all the more unbelievable.

As you recall, they never made it to the airport, or they made it to the airport, they had to execute a go-around and they never made it back again. It crashed on Long Island. The reason that they crashed is something that experts have never been able to agree on. It’s a point of contention in the airline community today that there’s this vast difference of opinion of who is to blame. Nobody can quite agree on who bears the blame for this crash because it’s a really unusual one. We’ll go through it from beginning to end.

They were delayed for quite a long time before landing, and this was the main problem. There were a lot of aircraft trying to land at New York that night so they had 39 aircraft in front of them. They were placed in to three different holding patterns, going around in circles at different points off the east coast of the United States, working their way up to New York. In the end, they were delayed for 77 minutes, so a very, very long delay. As they were circling in the holding patterns, there’s some very interesting dialogue between them and the controllers in New York. They requested a priority status to be given to them. Klotz, the first officer, he had the best English so he took care of all the communications to the American controllers. All the conversations he had with his colleagues in the cockpit were in Spanish. That was part of the problem.

He says, ‘Avianca 052, heavy. Expect further clearance time. I think we need priority. We’re passing … ’ What he says next is unintelligible unfortunately because at that point, he almost certainly makes a reference to the amount of fuel they have. It cuts out on the recording so we’ll never know what he said. The controller says, ‘Avianca 052, heavy. Roger. How long can you hold and what is your alternate?’ Meaning, what is your alternate airport? Klotz says, ‘Yes, sir. We’ll be able to hold about five minutes. That’s all we can do.’ The controller says, ‘Avianca 052, heavy. Roger. What is your alternate?’ He’s just repeating the same question. He doesn’t seem to have absorbed that information at all. Klotz says, ‘It is Boston but we can’t do it now. We will run out of fuel now.’

Prior to this, they hadn’t had much communication with the United States. Earlier in the flight, they had failed to call ahead and ask about the weather. There was that early problem. That was the first mistake in the process. At this point, Klotz is telling the controllers that they’re low on fuel. They can’t make their alternate airport, which is Boston, which of course is very, very close to New York. They’re obviously in a fairly dire position.

One of the controllers then telephoned through to Kennedy Airport in New York, asking if he can hand Avianca 052 over to them to take control of the flight. He says to Kennedy, ‘Can you take him or shall I offer him his alternate?’ It’s a very odd piece of communication. He hasn’t absorbed that piece of information that they can’t make their alternate airport. They can’t make Boston. They’re low on fuel. He doesn’t communicate to Kennedy that they’re low on fuel, and he creates the impression that they’re not low on fuel by saying, ‘Can I offer him his alternate? Should I offer him his alternate?’ Kennedy says, ‘No. We can take him.’ He’s handed over to … he says, ‘Avianca 052, heavy, cleared to Kennedy Airport.’

At that point, the communication is handed over from those controllers to the Kennedy controllers, who do not have that information about the low fuel. It becomes clear in later communications that Klotz is under the impression that they do have that information. Anyway, there’s a rather poignant piece of dialogue inside the cockpit now where Klotz says, ‘Right now, we are proceeding to the airport inbound and we have 27 – 17 miles. That means we’ll have hamburger tonight.’

There’s a bit of back and forth with him and the captain and the flight engineer where the flight engineer says, ‘They got us. They’re already vectoring us.’ Klotz says, ‘They accommodate us ahead of – ’ The captain interrupts him and says, ‘What?’ Klotz says, ‘They accommodate us.’ The flight engineers says, ‘They already know that we are in a bad condition.’ The captain says, ‘No, they are descending us.’ There’s a bit of disagreement. The captain seems to have some awareness that they’re not really being given priority. They’re just being descended at that point. Klotz then says, ‘They are giving us priority.’ The captain and the first officer are talking past each other and this happens quite a lot during the transcript. They’re not really communicating the information or agreeing on the information that they’re giving to each other.

At New York, the weather was absolutely terrible. There was wind shear high up and there was also wind shear very low down to the ground. Wind shear is where the direction of the wind changes rapidly. As you descend through it, the wind is coming from one direction then it wheels around to come from another direction. The Kennedy controllers informed Avianca 052 about the high-level wind shear, but nobody informs them about the low-level wind shear. They don’t know that they’re going to encounter wind shear just before they get to ground level, which is another mistake. All these mistakes are ticking up and adding to one another.

As they come in for their approach, there’s a vertical wind that pushes. It seems to push the plane downwards so they go down below the course that they should be taking, which is the glide slope, which is the correct approach to an airport. They’re being pushed down below that glide slope. There’s a great deal of confusion in the cockpit at this point. Interestingly, the captain seems to tune out and ignore all alerts about the glide slope. Klotz says to him, ‘Slightly low glide slope.’ Five seconds later, he says, ‘Below glide slope.’ Eight seconds after that, he says, ‘Glide slope.’ The captain doesn’t respond at all. A few seconds later, Klotz says, ‘This is the wind shear.’ He’s probably responding to the plane being pushed down and going under its glide slope. The flight engineer then chips in and says, ‘Glide slope.’

The plane then signals an alert and says, ‘Whip, whip. Pull up.’ Literally the plane says that over its speakers. ‘Whip, whip. Pull up.’ Klotz says, ‘Sink rate,’ alerting the captain. The plane then repeats that alarm, ‘Pull up,’ three times and then another four times and then another three times over the course of the next 15 seconds.

The captain then says something for the first time. He says, ‘The runway, where is it?’ The plane repeats two more glide slope warnings. Klotz says, ‘I don’t see it. I don’t see it.’ The captain says, ‘Give me landing gear up, landing gear up.’ He’s made the decision at that point, we can’t see the runway so we’ve got to do a go-around. They can’t see the runway because they’re further away from the runway than they think. Because they’re below the glide slop, they’re getting down that critical altitude where they have to make a decision far too early. The visibility isn’t good enough for them to see the runway, which is a kilometre or so ahead of them at that point.’

There’s two more glide slope warnings from the aeroplane, and then the captain says, ‘Request another traffic pattern.’ Someone radios the controllers and says, ‘Executing a missed approach. Avianca 052, heavy.’ There’s now an interesting piece of dialogue which is a behaviour that humans seem to have under high stress situations where they’re not focusing on how to solve the problem but they go over the problem that’s just happened where the tower says to them, ‘Climb and maintain 2000. Turn left heading 180.’ The captain says, ‘We don’t have the fuel,’ seemingly to himself. Klotz says to the tower, ‘Maintain 2000 feet. 180 on the heading.’ The captain says, ‘I don’t know what happened with the runway. I didn’t see it.’ The flight engineer says, ‘I didn’t see it.’ Klotz says, ‘I didn’t see it.’ There’s six seconds where no one says anything.

They’re going over the fact that they missed their landing, but they’re not talking about how they’re going to solve that problem. They’re dwelling on something that’s in the past and can’t be fixed, which is an unfortunate behaviour to have in a situation like that.

A big mistake happens. This is something that a lot of people think is the biggest mistake of all. The captain says – because remember the captain’s English is not good. He’s relying on Klotz to talk to the tower. He says, ‘Tell them we are in an emergency.’ Klotz then radios to the tower and says, ‘That’s right to 180 on the heading and we’ll try once again. We’re running out of fuel.’ He says, ‘We’re running out of fuel,’ rather than ‘We are in an emergency.’ There’s possibly a linguistic problem here in that apparently in Spanish, the phrase for requesting priority is a more urgent one than it is in English. Some people suggested that perhaps Klotz is thinking with his Spanish brain and thinking that he’s phrasing things at much more of an urgent way than he is.

Malcolm Gladwell has an interesting theory where he talks about Colombia as being a society with a great reluctance for junior people to assert themselves to more senior people. It’s a society where very subtle inflexions of voice and phrases are used to carry more meaning than they are in English. We’re much more ready to say if we’re in an emergency, we’re in an emergency, or if we need help, we need help, even if we’re talking to someone more senior than ourselves. That’s an interesting hypothesis, but I don’t know. I don’t know whether that’s true. That’s a hypothesis that Malcolm Gladwell put in his book. I think it was in Blink. He based that on work of a psychologist called Hofstede, if you’re interested in chasing that up.

Anyway, the tower responds and says ‘Okay.’ The captain says to Klotz, ‘What did he say?’ Klotz says to the tower, ‘Maintain 2000 feet. 1-8 on the heading.’ He says to the captain, ‘I already advised him that we are going to attempt again because now we can’t,’ and he doesn’t complete that sentence. The captain after a three-second delay says to him, ‘Advise him we are in an emergency.’ Four seconds after that says, ‘Did you tell him?’ Klotz says, ‘Yes sir, I already advised him,’ which he did not. At no point has he said ‘We’re in an emergency.’ That’s a problem.

The tower at this point hands them back to the approach controllers, which is a separate set of controllers. The controllers tell them what vectors they need to follow in order to re-land at the airport, but they sent them out in a very, very big circle, much larger circle than should have happened.

At this point, one of the air stewards comes into the cockpit. We know this because this air steward survived. She came into the cockpit to see what was the matter and asked the flight engineer who sits the furthest back – asked them what was going on, presumably ‘Why did we miss the airport?’ Apparently, she said the flight engineer didn’t answer her. He just looked up at her and just made that sign. The flight engineer knows that that was their only attempt and they don’t have enough fuel to get around again, but weirdly, the flight engineer has not said that to the captain in that first attempt. There were lots of talk about how to conserve fuel on that attempt, but at no point did he say ‘This is our one chance. We’re not going to get another chance after this.’

QUESTION: [inaudible] … that dichotomy you were talking about in levels of power.

OLIVER MACGREGOR: Yes, that’s right. This is what Hofstede certainly thinks is happening, that Klotz and Moyano are extremely reluctant to say anything to the captain that might be construed as critical or assertive or trying to do his job for him. The captain at the same time is not asking them for information. He’s not doing a very good job of leading and drawing upon their resources. The whole system between the three of them is non-functional.

QUESTION: But again it might be that if the lower parts of a hierarchy are so reluctant to say anything to the higher parts of a hierarchy, then maybe the higher parts of a hierarchy are reluctant to show that they don’t know what they’re doing.

OLIVER MACGREGOR: Yes. I think that’s quite right, Yes. I think if that assessment of Colombian culture is correct, yes, you would expect that mechanism to occur, that you’ve got that great power distance between the person who’s in a role of leadership, and so the person in the role of leadership feels insecure about asking for help because they don’t want to show that they’re not in control. They’re not on top of things.

Anyway, there’s a bit more dialogue where Klotz could have communicated to the controllers but he fails to do so. He says, ‘Avianca 052, heavy. We just missed approach and we’re maintaining 2005.’ The approach controller says, ‘Good evening. Climb and maintain 3000.’ The captain says to Klotz, ‘Advise him we don’t have fuel.’ Klotz says, ‘Climb and maintain 3000 and we’re running out of fuel, sir.’ He says this in the most calm tone of voice imaginable. It’s eerie to listen to. He literally says it. He says, ‘And ah, we’re running out of fuel, sir,’ like that. There’s no urgency in his voice at all. It’s very odd.

Eighteen seconds after that, after a little bit of back and forth just about technical details of where the plane is heading, the direction, the captain says, ‘Did you already advise him that we don’t have fuel?’ Klotz says, ‘Yes, sir. I already advised him. 180 on the heading. We are going to maintain 3000 feet and he’s going to get us back.’ The captain says okay. He’s getting them back, but he’s getting them back in far too wide a circle. They’re going a long way from the airport at this point.

The approach says – and this is the last critical bit of communication between the approach and the plane that deals with this issue – he says, ‘And Avianca 052, heavy. I’m going to bring you about 15 miles northeast and then turn you back on to the approach. Is that fine with you and your fuel?’ Klotz says, ‘I guess so. Thank you very much.’ 15 miles and he says, ‘I guess so. Thank you very much,’ again in that tone of voice.

At this point, we’re almost at the point of no return. There’s not much time left to go. There’s a few more critical pieces of dialogue where the approach says, ‘Avianca 52, climb and maintain 3000.’ Klotz responds, ‘Negative, sir. We’re just running out of fuel. We’re okay, 3000, now okay.’ There’s some urgency coming into the situation now that he’s responding to the approach each time saying, ‘We’re running out of fuel.’ There’s clearly a lot of confusion in his mind as to why that message isn’t making more of an impact with the approach, given that presumably he thinks that the approach has been briefed by the other controllers how critically low they were.

Approach says, ‘Okay. You’re number two for the approach. I just have to give you enough room so that you can make it without having to come out again.’ They’re number two in the approach. They’re putting other planes ahead of them at this point. They’re not giving them priority at all.

That is just a few seconds before the flight engineer says, ‘Flame out. Flame out on engine number four.’ That means that engine is doing a little explody thing where the last bit of fuel is being puffed out and it’s bursting into flame. A couple of seconds later, he says, ‘Flame out on it. Flame out on engine number three. Essential on number two. One, number one.’ The captain says, ‘Show me the runway.’ He thinks we must be close to the runway. He doesn’t realise at this point they’re ten kilometres away from the runway. Nobody knows who says this because they can’t work out the voice, but they’re assuming it’s Klotz who says, ‘Avianca 052. We just lost two engines and we need priority, please.’ Two seconds after that, the recording goes dead. Then there’s that emergency call to the police from the lady in Oyster Bay whose yard got a little bit messed up.

That’s the situation. That’s the story of Avianca 52. It’s an absolutely fascinating one. The more you delve into it, the more interesting details there are, but it’s a classic example of a disaster that builds up very, very, very gradually. There’s no discernible single cause. There’s a lot of very small events that overtake the people involved. The disaster happens around them without any of them becoming aware that it’s gradually overtaking them and surrounding them. That’s how it occurs. At no point does anyone really make a positive error. It’s more just a number of errors of an absence of doing something. There’s dozens and dozens of points where you think if only someone had done this or done that or done that, the whole process would have been short-circuited and it would have been averted.

I wish that people who describe themselves as safety officers, all these safety people that we now have to deal with in all walks of life, I wish that they would study this crash just to remind themselves that disasters are not necessarily straightforward and identifying blame is not always as straightforward as they often depict it.

QUESTION: The engineer had an unfortunate name.

OLIVER MACGREGOR: Yes, he did. Mauricio Klotz is a very ironic name for him to have had. Poor guy. The happy ending to this story is that after this crash and a number of others … when I first started researching airline crashes in the early 2000s, the airline industry is obsessed with this mechanism of breakdowns of communication. Throughout the [19]90s, they put a huge amount of effort into improving communication within airline cockpits and between planes and ground controllers and everyone else like that. They fairly confident they’ve fixed that problem now. The rate of airline crashes between then and now has gone down massively.

QUESTION: You were talking earlier about the cost of transporting stone from one place to another. I did a very short archaeology course many, many, many years ago. We were taken to what I look at as a factory site around the Merry Beach area on the South Coast. There’s an awful lot of flakes and discarded cores around the area. The problem is the stone doesn’t come from that area. Some of the stone comes from way up north or way down south. At the time, the person who took us said, ‘Obviously, they brought it here for religious purposes.’ I sort of took this a little bit sceptically.

I’m wondering whether you’ve got any thoughts on how that stone actually got there, why they would take that amount of effort to transport stone that far. I’m talking hundreds of miles, nearly a thousand miles – why they would put that effort into transporting the stone that far to make it in that place. As far as I know, they’ve never found the actual tools themselves. Obviously, the tools must have been traded back somewhere.

OLIVER MACGREGOR: Yes, potentially. There’s a lot of odd things that happen on the coast, along the east coast of Australia. Sometimes you have sites using very local materials. Sometimes you have people trading things over vast distances. Sometimes a great combination of the two. I think there’s a number of possibilities that explain it without having to invoke religious or spiritual mechanisms, which of course is always the default explanation when you can’t think of any other explanation in archaeology.

I think you could also explain it in terms of there’s clearly a lot of trade along the coast, a lot of commodities. There’s a lot of networks of tribes along the coast that survived up until European contact. It’s certainly not unusual for things to be traded over long distances in that part of Australia and other parts. It doesn’t have to be coming in one trip as well. That’s the other thing. You can have things being passed off from group to group. It’s always as soon as you find something traded, you immediately think, well, this is one great, big trade, expedition type thing. It could be a series of smaller trade networks where things are coming down gradually.

QUESTION: What I was actually wondering was that maybe this particular group that were around Merry Beach – wherever it was – were particularly good flint knappers and therefore they were trading because they got better tools that way, because maybe they’ve used the techniques that you were talking about. It was a sort of factory site because of the expertise.

OLIVER MACGREGOR: Possibly. Yes. I don’t know. I don’t know that anyone has proposed that, but it’s certainly a possibility that you could have certain groups where the level of craft expertise was higher, and so those groups might have been able to exert more of a control over the technological flow of materials. Possibly, yes. I don’t know.

QUESTION: If you’re going to trade, wouldn’t it make more sense to trade an already knapped piece of stone? Because it’s already done. It’s lightweight to transport. Wouldn’t that make more sense?

QUESTION: That sort of says it’s really good motivation for transporting the un-knapped stone, doesn’t it?

QUESTION: Yes, because you think that would be so much heavier to transport.

QUESTION: It’s sort of like sending [inaudible]. Trying to schedule a [inaudible].

OLIVER MACGREGOR: That’s right. Exactly. Yes. Who knows? The problem with trade networks is we see the tangible parts of it, the durable materials like stone, but the frustrating bit is we don’t see all the intangible things that went along with it. What else was being traded? What other materials that don’t survive in the archaeological record were making their way up and down the coast? Could the stone actually just be a completely incidental commodity that people are taking along with them and embedding in other trips and expeditions and movement up and down that coast?

QUESTION: I was going to say what was so special about the particular stone that was being used at that site? Was it some special kind of stone?

QUESTION: I don’t remember that actually. I suspect that Merry Beach has – correct me if I’m wrong, but the stone around Merry Beach isn’t particularly good for knapping. I think whatever stone –

OLIVER MACGREGOR: Good question.

QUESTION: – was that they were importing was a really good flint that was good for knapping, and they didn’t have it. Then we go back to why not just import what’s fully knapped flints?

QUESTION: They might have needed more goods in exchange for a finished product.

QUESTION: Ah, that’s a possibility. Yes, I agree.

QUESTION: I’m just wondering. For someone like myself who knows nothing about knapped stones but I often go bushwalking and kick the odd stone and have a close look at it; are there any discerning features you should be looking at to know, what is a knapped stone versus rocks that’s got some chips in it? [laughter]

OLIVER MACGREGOR: Yes, there are. Hold on. I’ll just bring up those illustrations. [shows slide] The usual features that a diagnostic of a stone flake that is being produced through impact fracture, are you get the fracture initiation which is occurring at the impact point, so you get a very constricted starting off point of the fracture. Directly below that impact point, you get a conical fracture. This is the Hertzian cone I mentioned earlier. That’s the first little bit of the fracture that starts off at that impact point. It’s a cone that extends into the material underneath that and then the fracture propagates out from that. Because it starts from that cone shape, it propagates outwards and then recurves back on to itself so it produces a bulb-like cross section. I’ll put this up full size.

In cross section, that’s what it looks like. We’ve got the cone at the top. The fracture propagates from the inside edge of the cone. It starts off at that angle but then it recurves back towards the free surface of the core. You get a bulb shaped thing. This is happening in three dimensions. You get this hemispherical shape to the fracture surface. That’s the most diagnostic feature. If you have a piece of rock that’s reasonably fine-grained and it has that visible on it then it’s almost certainly –

QUESTION: That would be a symptom of a tool?

OLIVER MACGREGOR: It is at the platform surface, so this is all happening at one edge, it’s all happening at the platform edge, so you’ve got the fracture focalising up at that impact point and then the bulb is directly below that. I have got a photo in here somewhere. [shows slide] That’s looking straight down on to the ventral surface, on to the fracture surface that’s produced through that impact fracture. In this case, it’s been hit up there, and we’ve got the bulb occurring underneath it there, and then the fracture propagates down from that point. You usually get ripply type features in the stone as the fracture is deviating back and forth like that. They take the form of ripples because the fracture is propagating out from that point so it’s propagating out in a circular fracture front. Each time it wiggles and wobbles up and down, it produces these semicircular ripples in the fracture.

QUESTION: It’s almost like a stone hitting water.

OLIVER MACGREGOR: Yes, that’s right. A lot of people have compared it to that. Yes. It’s like ripples through a pond.

QUESTION: Another question, sorry. How common are they when bushwalking around the South Coast or whatever.

OLIVER MACGREGOR: Very, very common.

QUESTION: Very? Worth looking for?

OLIVER MACGREGOR: Yes. It varies from place to place. Yes, that’s right. Obviously they all have legal protection and all that stuff. Technically when you find them, you should report them and fill out site cards and that thing. Following the –

QUESTION: You’re not allowed to move them.

OLIVER MACGREGOR: Not allowed to move them, no. That’s right.

QUESTION: … once you return, you are not allowed to move them.

OLIVER MACGREGOR: That’s it. Exactly. It used to be the case in New South Wales – I think it still is that if you’re an archaeological professional – so if you work as an archaeologist and you find a stone artefact, I think you’re legally bound to fill out a site card and record that as a new site. Is that right? No one is going to contradict me? Let’s say that’s right. Actually for people like me, there’s a legal imperative that if we find something when we’re bushwalking, we need to go back and fill out the paperwork online and lodge it with the government as a new site.

QUESTION: You’ve given us a very hard to put explanation of flaking. What you haven’t said is the different varieties of rock can flake in a different way. You will have a flint that knaps in one way, but then you might have a completely different type of rock which is also used as a cutting implement, knaps in a different way, because of the way the rock or geology of the rock itself. You talked about glass which is a fluid. Some rocks are fluid. Some have a different basis. So not all knapping is for [inaudible]

OLIVER MACGREGOR: Glass is a solid rather than a fluid. If you’re talking about fluids in terms of –

QUESTION: [inaudible] You break the viscous and therefore you –

QUESTION: Glass will actually flow. It is actually a fluid and it will flow.

OLIVER MACGREGOR: No.

QUESTION: If you go to Questacon, they’ve got an example of it.

OLIVER MACGREGOR: That’s actually a myth. I know.

QUESTION: Oh, really?

OLIVER MACGREGOR: Yes. We were all taught that in high school, and I was a bit shattered to learn that this is not the case, but yes, glass actually isn’t a fluid and it doesn’t flow over time. They say that old windows become thicker at the bottom than at the top, but when they look at old windows, they find that as often as not, the windows are thick at the bottom than at the bottom. It just seems to be that when people are making glass in the olden days, they couldn’t make flat plates with the precision that we make today so you’d have bits that were irregular in their thickness.

QUESTION: My apologies.

OLIVER MACGREGOR: There were a number of physicists who did – first of all, there was one noble physicist who decided to test this. He got a glass rod and he sat it on two supports at either end and then he put this great heavy load in the middle so that it bent like a horseshoe, and then he left that sitting there for 30 years [laughter]. He started this experiment in the [19]60s and presumably just had it sitting in his garage for all that time with strict instructions to his family not to move it under any circumstances. He took the weight off at the end and there was no measurable deflection, so there’s no plastic defamation of the rod in that 30-year period.

There’s a bunch of other physicists who did some very high level theorising that I don’t understand at all. They worked out that if glass is a liquid, if it were a liquid, its molecules would be moving so slowly that it would take more than the length of the universe for any observable movement of those molecules relative to each other to occur. They’re really fun papers. I reference them in my PhD. I didn’t understand them at all, but in the conclusion – I didn’t understand the main body but I understood the conclusion bit which said, ‘This means that glass is not a liquid.’ [laughter]

The other thing of course is that we have glass artefacts from ancient Rome and ancient Greece and time periods like that. If glass was moving over any human time scale, those wouldn’t exist anymore. So, glass is a solid.

Returning to the original question: different materials do fracture in slightly different ways, yes, but the differences are fairly subtle. The fundamental properties of fracture of all different materials are more or less the same. These sorts of gross differences that we observe in one material are likely to carry over into others as well.

QUESTION: Did knappers use boulders as anvils or as vices?

OLIVER MACGREGOR: Yes, they did. People would flake rock holding it on an anvil. Rather than just flaking them in their hands, they occasionally placed them on an anvil and flaked them that way. That’s ‘bipolar knapping’. That does occur. That’s a good mechanism for increasing the inertia of the core and therefore, being able to apply more force to small pieces. Peter Hiscock did a good paper on that in the [19]80s I think it was where he identified cores on archaeological sites that were decreasing to a point where most of them get discarded but then some of them get retained for quite a long period after that. He said that was the threshold at which people are switching from freehand knapping to bipolar knapping. I think that’s in AA [Australian Archaeology] somewhere. That’s another strategy, essentially achieving much the same aim.

QUESTION: How sharp can they be?

OLIVER MACGREGOR: Stone artefacts can be very sharp. Theoretically, the edge of a stone artefact could go right down to one molecule of thickness. There was a paper by some surgeons where they tested the cutting power of flakes made out of obsidian, which is a volcanic glass, a very, very fine grain. It’s an amorphous material just like manmade glass. They tested the cutting power of those compared with surgical scalpels, and they found that the obsidian was actually sharper. They have been used in a lot of high-level surgery. They were being used in eye surgeries a few times I think.

I think I read somewhere that actually they were a bit too sharp for some surgeries, and in some cases, they inhibited the wound healing because the cut was too clean or something like that, but I’m not sure. I can’t remember whether I read that in a reputable source or one of those amateur archaeology magazines, or something like that. In theory, they can be very, very sharp. Even if you don’t have access to obsidian, even the more working A materials that we commonly find around here, you can get a very sharp edge without any particular expertise at all, even on common materials like quartz and that sort of thing. You can certainly make something sharp enough to shave with.

Any further questions? All right, thank you very much, guys. It’s been a great pleasure. Ever since I watched my friend, Adrian Di Lello, give a CAS [Canberra Archaeological Society] talk in 2003, I’ve been thinking it would be a good thing to do. Finally a decade and a bit later, I finally got around to giving one. It’s been great. Thank you.

[applause]