Roman Glass

Alright, internet, I just finished writing a goddamn master’s thesis and I am gonna learn you some things that I only just got done learning myself about the chemistry of ancient Roman glass.  No, it has nothing to do with Pokémon, but it’s happening whether you like it or not, so buckle the f@&k up, kids.

Okay, first things first: someone’s probably told you at some point “glass is technically a liquid” followed by something about Mediaeval church windows.  LIES.  Pervasive lies, a myth people keep repeating, something you should eradicate from your head right now.  Glass is no more a liquid than rock.  Yes, Mediaeval window panes are often thicker at the bottom, as if the glass in them has flowed downward over the centuries.  They’re also often thicker at the top, or at the sites.  They’re not like that because the glass is actually moving; they’re like that because Mediaeval window glass is $#!t.  Now you know.

So what is glass, then?  It’s an amorphous solid.  The glass we use in windows and drinking glasses is mostly silica, the same stuff that sand and quartz are made of, but where the atoms of quartz are neatly arranged in a regular, symmetrical, repeating pattern (what we call a “crystalline” structure), the same atoms in glass are pretty much all over the place.  Any solid that lacks a proper crystal structure is a “glass.”  Plastics can be “glassy,” and so can metals, even.  Obsidian, a.k.a. volcanic glass, a.k.a. dragonglass, for fans of A Song of Ice and Fire?  That happens to be mostly silica as well, but the reason we call it glass is because it’s also amorphous, randomly structured, where most rocks are regular and crystalline (up to a point).  When you make glass, you’re basically taking sand (or sometimes crushed quartz pebbles) and melting it, which destroys the crystalline order of the solid, then cooling it down fast enough that there’s no time for those regular structures to reform before the molecules ‘freeze’ – so what you get is something like a ‘snapshot’ of molecules in the kind of shifting chaos that’s normal for liquids.  That’s probably where the whole “glass is really a liquid” thing comes from; its molecular structure does kind of resemble one in some ways.

Now, melting sand in the first place is really difficult, because pure silica has a melting point of about 1700 ºC (hard to say what the hottest Roman furnaces could do, but I’d guess 1300 or 1400 ºC at the absolute most).  You need to mix it with what’s called a “flux,” something that lowers its melting point and also makes it more fluid once it melts, so it’s easier to work with.  This happens because atoms from the flux get mixed up with the atoms from the original silica lattice and get in the way, making it harder for crystals to form – something similar happens when you scatter salt on the roads in cold weather, to lower the melting point of ice.  The flux used in Roman glass is a substance called natron (nitrum in Latin), which is a mixture of sodium compounds that you dig up out of dry lake beds in Egypt (the ancient Egyptians used the same stuff to dry out corpses for mummification).  This is actually quite odd; most historical glasses are made with the ashes of plants, particularly plants that grow well in salty water, because those contain a lot of sodium.  Roman glass actually resembles the glass we use today for windows, drinking vessels and the like much more closely than the glass that was made in, say, ancient Egypt, or Mediaeval Europe.  Using natron gives a purer, better quality glass than plant ash, but natron is a bit more difficult to come by because there are only a few places in the Mediterranean world where you can gather it up in large quantities; besides Egypt, the only major source we know of at this point is Lake Van, in eastern Turkey.  Glass also contains calcium, which improves its resistance to corrosion, and you can get this either from the plant ash (which is made of a whole mess of different chemicals) or from crushed seashells (made of calcium carbonate) in the sand.  Glass with no calcium will gradually dissolve in water, which is what has happened to a lot of early Iron Age glass from Egypt and Mesopotamia – after they started using natron, but before they stopped using crushed quartz (which is purer than sand, but has no calcium in it).  Roman glass also normally contains small amounts of iron and aluminium, which come from the sand, and traces of a whole bunch of other elements, notably titanium, zirconium, and chromium.  All of these things are mixed into the silica lattice and chemically bonded together in one great big mess, almost like how atoms of different metals are mixed together in an alloy.

There’s a whole bunch of things you can throw in with the molten silica to make coloured glass.  The iron in the sand makes most Roman glasses naturally a sort of blue-green colour, and actually this is the colour you see when you look at a pane of glass edge-on – we use much more pure ingredients in glass today, so the colour of the iron is too weak to notice from most angles, but it’s there.  You can create a range of other colours from green to amber without adding anything extra at all by manipulating the furnace atmosphere to change the oxidation state of the iron (this actually works by causing it to react with traces of sulphur in the glass).  Adding cobalt compounds makes a deep blue glass, copper makes more of a turquoise blue (or opaque red, depending on the oxidation state), manganese makes purple, calcium antimonate makes an opaque white, and lead antimonate gives you an opaque yellow.  Mixtures of some of these are also possible, and adding certain manganese or antimony compounds in just the right amounts will ‘neutralise’ the natural blue-green colour of the iron, leaving you with a colourless glass.  Most of these ‘recipes’ had been around for ages before the Roman Empire got started; lots of them were used in glassmaking by the Egyptians and Babylonians way back forever ago, and continued to be used centuries after Rome ‘fell’ (whatever that means, and whenever you believe it happened).

Now here’s the interesting part.  As you might have guessed from my incessant rambling about chemical compositions, ‘scientific’ approaches to archaeology are kind of my schtick.  For a good 60 years, there has been a notion in archaeology that the particular chemical composition of a piece of pottery can be used as a kind of chemical ‘fingerprint.’  If you match that up with the composition of material from a clay deposit, you can figure out where the raw materials came from, and therefore where the pot was made – which might be a completely different region from where you found it.  That’s exciting, because it means you can track the movement of artefacts (which means trade), and maybe people (which means migration).  It’s a little more complicated than that, for a lot of reasons.  Sometimes pottery doesn’t have the same composition as the clay it was made from, because it’s been ground and washed to make it finer, or because it’s had gravel or something added to it to make it easier to work with.  Besides that, there are a lot of ways that an object can get from one place to another, especially in a fairly interconnected and cosmopolitan civilisation like the Roman Empire, and it takes a lot of samples to figure out meaningful patterns, but when you can get this sort of thing to work, the results are often really interesting and tell you some neat things about how particular sites fit into big regional economic patterns.  Now, in theory, you can do exactly the same thing with glass, looking to match it up with sources of sand, and there’s been a lot of interesting work in that area in the last 20 years.

Here’s the thing, though: most Roman glass (or at least, glass that hasn’t had extra ingredients added to colour it) seems to be pretty much the same.  Same ratio of silicon to calcium to sodium, same amounts of iron and aluminium, very little of anything else, which suggests that the sand used to make it is very clean, very pure.  It’s a dead ringer for the sand you get in parts of the Levant (the coastal region that includes Israel-Palestine, Syria, and Lebanon – you might be familiar with the name from recent news stories about our dear friends, ISIL).  Based on the more straightforward forms of chemical analysis available, it seems like pretty much all Roman glass was made in that relatively small section of the Empire.  Weird, right?  Especially since we know that there were glassblowing workshops all over the place during the Imperial period – we’ve found places like this in Egypt, Tunisia, Italy, France, Belgium, even Britain, which is about as much of a backwater as you can possibly get, as far as the Romans are concerned.  What’s going on?

Well, when you think about it, there are really two stages in making a glass object – first you have to make the glass from sand and natron, and then you have to get an artisan to turn a shapeless lump of glass into something you can actually use.  It seems like most Roman glass was originally made in the Levant.  We have archaeological remains of only one or two places where they did this; they seem to have used these really huge melting tanks to produce enormous slabs of glass, which then basically had to be mined out of the furnace with a pickaxe or something.  We actually have one glass slab that went wrong because it had the wrong proportions of ingredients go into it, and so was never broken up to be sold and used.  It’s in the middle of a disused Jewish cemetery in a place in Israel called Bet She’arim, it was made at some point between the 4th and 9th centuries AD (which is annoyingly vague; it may well be Islamic, not Roman), it’s about two metres wide, four metres long, and half a metre thick, and it weighs eight tonnes.  You can see why no-one’s ever tried to move the bloody thing.  We’ve never found anything like that in the western half of the Roman Empire, but what we have found is Roman shipwrecks carrying rough, irregular chunks of glass, exactly what you would get after breaking up one of those huge slabs; it seems like they were being shipped across the Empire so glassblowers everywhere could use them.  So it seems like the archaeological evidence fits with the chemical evidence.  Case closed, right?

…well, no; it’s never that bloody simple.  See, the thing is, my good bro Pliny the Elder, who wrote the Latin encyclopaedia known as the Natural History, talks about glass in his chapter on stones and architecture (odd place to put it, you might think, but in the ancient world glass seems to have been considered a kind of artificial stone – which, in fairness to them, is not an unreasonable description from a chemical perspective).  He describes the great big furnaces in Levantine cities like Tyre and Sidon – and he also says that the Romans have recently (he’s writing in the mid-1st century AD) started building them in Italy, France and Spain.  Hrrrrm.  Now, Pliny says a number of ridiculous things – remind me to tell you all some time about the centaur preserved in honey – but I have trouble believing that he could be wrong about this.  He might be confusing a glassblowing operation (using imported raw glass) with a glassmaking operation (using the big melting tanks), but then again, he specifically mentions that the Roman glassmakers use sand from the mouth of the River Volturnus, in Campania (the region of Italy where Pompeii is).  For one thing, that kind of specificity would be really bizarre if he were getting mixed up by an operation that actually wouldn’t have used sand at all.  For another thing, Campania is where Pliny lived for at least part of the year.  He probably never visited the facilities personally, but it’s not like he’s going off some third- or fourth-hand report about how people do things at the other end of the Empire; this is, like, a couple of towns over from his country house in Misenum.  Still, with no archaeological evidence, it’s relatively easy to shelf him on this one, since Pliny is, on the whole, marginally less reliable than the Pokédex.

But – of course there’s a ‘but’ – remember what I said about how this was all based on “the more straightforward forms of chemical analysis available”?  Well.  It turns out that when you start looking at Roman glass using some more esoteric stuff – in particular, neodymium isotope analysis, which is still quite new on the archaeological scene, and has only been in use for about ten years – things get… less clear.  When you do stable isotope analysis, you’re looking at just one element in your sample (in this case, neodymium, which is present in tiny amounts in most sands) and measuring the amounts of different isotopes that it contains – atoms that have different numbers of neutrons, and therefore different atomic weights.  Isotope studies can give you a chemical ‘fingerprint’ in the same way as trace element studies – a signature of sorts, that you can match up with regional geology.  In the case of neodymium, it turns out that sand from the Levant contains neodymium that is just a little bit ‘heavier’ than neodymium from sand in the western Mediterranean – this is because Levantine sand comes from the Nile, and ultimately from volcanic sediments in east Africa, while west Mediterranean sand comes mainly from wind-blown sediments from the Sahara desert.  The practical upshot of all this is that, if you look at its neodymium isotope composition, you can tell whether glass was made from west Mediterranean sand or east Mediterranean sand.  And guess what?  In collections of Roman glass, there’s some of each.  That means that, just like Pliny says, there are probably primary-level glassmaking facilities at both ends of the Roman Empire.  What about the other analyses, the ones that say the glass is all basically the same?  Well, there are a couple of long-winded explanations for that, but in short, it seems like maybe the glass all winds up looking very similar because of some aspect of the way they make it, not because of the raw materials – possibly they have some way of purifying the sand before they start, for example (and Pliny, again, actually does seem to describe something along those lines; he talks about grinding it with a mortar and pestle, which would strip out a lot of the softer mineral impurities).

Most of the actual legwork on projects like this tends to be done by geochemists, not by archaeologists or historians, and a lot of the time geochemists have only the vaguest idea of what archaeologists actually do and what we care about.  Meanwhile, a lot of archaeologists, particularly in Greek and Roman archaeology, aren’t terribly familiar with how chemical analysis works, and aren’t keen to learn.  So the result is that some geochemist spits out this report about where glass was made in the Roman Empire, the archaeologists shrug and say “okay, but what does this tell us about peoples’ lives, beliefs, society, culture, that sort of thing?”, the geochemist shrugs back and says “I’unno,” and the whole thing winds up being just a really expensive pile of nothing that no one pays much attention to.  It doesn’t help that very few archaeologists are interested in glass specifically – lots of people study pottery, lots of people study art and sculpture, lots of people study bones… but the people who study Roman glass, quite often, are people who study the history of glass in general, and those people don’t talk to Roman archaeologists all that much.  And that, to me, is a very unfortunate thing, because come on!  The Romans took what they saw in the Levant and turned it into an industry that spanned all of Europe.  They had high quality glass being made all over the Empire, apparently at a fraction of the cost compared to earlier periods.  Their methods for making glass were apparently standardised to a degree that we can’t tell the difference between Italian and Levantine glass without ridiculously specific analytical methods.  And let’s not forget that they would have needed to get huge quantities of natron – which is basically only produced in Egypt – to wherever glass was being made, on a regular basis, for dirt cheap.  “How ‘modern’ was the Roman Empire?” is a really big, important question for people who study ancient economics and society, and the glass industry, I think, is easily one of the most modern things about the Romans.  If we just sit down and think through this stuff, that’s got to be important.

So basically my MA thesis is a vague attempt to make Roman archaeologists shut up and pay attention to all this nonsense that I’ve been ramming down your throats!  If you’ve been paying close attention, you are now better informed about this one unbelievably specific aspect of Roman civilisation than most people who study Romans for a living!  Good job!

Now, just for fun.  Two of my favourite things that I’ve learned about Roman glass are things that are completely tangential to what I was actually writing about in my thesis, but are also absolutely awesome.  I really think I have to share these with you.

First thing.  Remember the stuff about different fluxes you can use in glass?  Well, they’re pretty easy to tell apart, chemically.  Glasses made from the ashes of saltwater plants are normally high in magnesium and phosphorus as well as sodium.  In Mediaeval Europe, you also get glass made from wood ash, which is quite different – trees do not grow on salty water, so these are low in sodium, but instead high in potassium, as well as magnesium and phosphorus.  So… what happens when you come across a glass that’s high in potassium and phosphorus, but low in magnesium, and has no sodium at all?  Well, if you’re like me, you literally type the word “what” into your notes several times over before you continue reading the article.  We’re talking here about some samples of glass from sites that belonged to my old friends the Etruscans.  They are a group of really quite nondescript blue beads, and they’ve been knocking around private collections long enough that we don’t actually know where they came from, beyond “somewhere in Etruria” (they may originally have been looted; lots of Etruscan stuff was), which makes them practically useless for archaeological purposes.  This is really frustrating, because it means that we can’t say much at all about the implications of that really interesting composition, and I would like nothing more than to spend a couple of years looking for more glass objects like these with a more secure archaeological source.  See… low magnesium means they weren’t made with the ash of any sort of plant, but high calcium and phosphorus… what’s made of calcium phosphate?

Go on.  Google it.  I’ll wait.

Did you do it?  Yeah?  That’s right – bone is made of calcium phosphate.  This glass was probably made from bone ash.  As far as I know, this is unique, not just in the ancient Mediterranean, but in all of history.  And if it’s true, well, that’s really interesting, because when do you get bone ash in the ancient world?  Answer: animal sacrifice.  Animal sacrifices were a big deal for the Greeks and Romans; they were huge communal meals, which for a lot of lower class people were the only time they ever ate meat.  And if someone was making glass from the ash left over by these really important religious and cultural events?  That has to say something interesting.  I just don’t know what, because there are so few damn samples, and their history is so poorly attested!  Argh.

In short, there might just be such a thing as Etruscan bone glass, which is the single most badass phrase I have ever had the pleasure of typing in a semi-serious piece of work.

The second thing has a story behind it.  See, one day when I was starting my research, I was searching for articles on JSTOR, which is a big online database of academic publications from lots of disciplines, mostly in the humanities.  Now, the thing you have to know about JSTOR is that JSTOR’s search engine is completely bat$#!t.  I often have to go through two or three versions of my search terms to get it to spit out anything related to what I want to know (in an unrelated incident, it once put me onto a 1919 article by a man who believed, apparently with absolute sincerity, that window glass would lead to the evolutionary downfall of the human species).  On this particular day, I was trying to learn about the different colours of glass used in the Roman world, and typed in “Roman coloured glass.”  Of course, it makes perfect sense that the first result would be a 1948 article entitled “The Earliest Known Use of a Material Containing Uranium.”

Now, just for a moment, put yourself in my position.  There was, I am sure you will agree, no way in the space of all human potentiality for me not to read that article in its entirety.

To my absolute joy, this article (written by a man, Earle Caley, who was one of the great pioneers of applying chemical analysis to archaeological materials) really was about Roman coloured glass.  Caley was commenting on the results of an even older study (1911, if memory serves), which had detected, in a glass mosaic tile from Roman Italy, a uranium dioxide content of some 2.5% (which might sound low, but it’s way too high to be accidental; trust me on this – some colourants, like cobalt, you only need about 0.1% to get the effect you want).  This is not as weird as it sounds, because actually, in the 19th century, it was really common to make glass with uranium – it gives the glass a quite distinctive lime green colour.  So what seems to have happened is some curious Roman maker of mosaic tiles has stumbled upon a source of uranium ore, probably pitchblende, and thrown it into the glass mixture to see what happened.  This is a curious thing to see happening, because the Romans, as far as we can tell, were not really in the habit of experimenting, of trying new things just to see what happened – they were, for the most part, traditionalists.  They did things that they knew worked.  If this analysis was accurate (which, let’s be fair, it may not have been, this was 1911), then it seems like we’re seeing a very unusual, very interesting approach to technology and innovation that’s quite different from the way we’re accustomed to thinking about the Romans.

Actually, the really weird thing about the glass isn’t even the uranium, it’s that the 1911 analysis reported no sodium, a little bit like the bone glass I was just talking about.  To be perfectly honest with you, I’m a little suspicious of the whole thing; I want to go and find the original publication, if I can, and check up on their methods.  In 1911, they wouldn’t have had fancy spectroscopic machines like we use today; they would have been using ‘wet chemical’ methods – physically dissolving some of the sample, using a series of reactions to sift out different bits of it, and then weighing them one at a time.  According to Caley, the authors actually didn’t explain what tests they used for uranium (scientific rigour in 1911 was sloppy sometimes), only that they tried everything they could think of, because they also thought it was really weird and didn’t want to accept the result at first.  He trusts them on this point, which I think is probably reasonable, but what I’d like to do is look up what the standard tests for uranium would have been in 1911 and try to figure out what, if anything, could have given them a false positive.  If I actually find something, maybe I can present it at some conference or other and really confuse everyone, or even publish it as a journal article.

So that’s me on Roman glass!  Hopefully I’ll be getting back to the stuff you people actually come here for pretty soon.  If for some ungodly reason you want to drag yourself through the awful $#!tstorm that is my actual MA thesis, which is two hundred pages long and dense as f@&k, it’ll probably become available online in .pdf form through my university’s website at some point in the near-to-medium future, or I’ll stick it on my page or something; whatever.  Peace out, b!tches.    

Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s