Select Page

I referred to this news story as being potentially the archaeological story of the decade on twitter. Potentially is a good weasel word, but if Rehydroxylation Dating can be independently verified then it could be a more important form of dating than radiocarbon dating. The reason it’s so exciting is that this method will allow archaeologists to date pottery. A couple of warnings before I start. I am not a materials scientist so it’s possible that if something seems odd that’s me messing up the description. The other is I am not on the research team – I’ve merely emailed some questions.


Saxon Pottery
Late Saxon Pottery, but how late? Photo (cc) Wessex Archaeology.

Pottery and other ceramics make up most of the data that you’ll find on an archaeological site. Unfortunately there hasn’t been an easy way to directly date it. The most common way is by style. Pot types and technology come into and out of fashion. Terra sigillata, Samian Ware, is particularly good for this as styles turned over rapidly. However, that no help if all you have is a fragment of cruddy Iron Age pot. Another method would be by association with organic material. If you find some grain in the same strata, you can date that and by association when material was deposited in that strata. There are some problems. Radiocarbon will give you a range of dates rather than one date. This range can be quite wide and it’s prone to contamination. What would be useful would be a way of dating ceramics directly. You can do this with thermoluminesence, which uses natural radioactivity to give a date, but it’s complex and difficult so it’s rarely used. A team mainly based at Manchester University have announced that they can date ceramic materials, such as pottery, tile and brick, through a process called rehydroxylation. It seems to be simpler than both thermoluminesence and radiocarbon dating and much harder to accidentally contaminate. There are some impressive additional uses for the method which could make a lot of excavated material a lot more useful.

How does it work?

The idea has developed from a known problem in architecture – kinetic expansion. Molecules in clay have sites which react with water, H2O, to take on hydroxyl groups (OH). When you fire clay to make a pot or a brick, you drive out these hydroxyl groups. Once you have your fired ceramic it starts reacting with water vapour in the atmosphere to take on hydroxyl groups again. The longer you leave it, the more OH the ceramic absorbs. Wilson et al looked at this in a paper from 2003, Kinetics of Moisture Expansion in Fired Clay Ceramics: A (Time)¼ Law (DOI: 10.1103/PhysRevLett.90.125503). They found that rehydroxylation of ceramics followed a power law, and that it was linear with respect to (time)¼.

powerlaw
An equation y=x¼ which flattens as you move right.
Plot by WolframAlpha.

In everyday terms it means that equal amounts of mass are taken up on a ratio of 1, 16, 81, 256… So if it takes a day (or a week, or a month) for a ceramic to increase by 1 gramme of mass then it will have increased by 2 grammes from its start weight after 16 days (weeks, months etc), 3 grammes after 81 days and so on. This gives you a graph a bit like the one to your right, which is a pig to read usefully, which is why Wilson et al plot against (time)¼ because that gives you a nice straight line.

After in their paper on kinetic expansion the authors mentioned the possibility of archaeological dating. Now they have a technique.

You take a sample around 3g to 5g in mass and heat it to 105ºC. You leave it here for a while to make sure the sample is dry and you’re not weighing any excess water. When you’re satisfied it’s dry, you weigh it to get the aged weight. Then you stick it in an oven at around 500ºC. This drives out the hydroxyl groups. You leave it to bake for four hours. When it’s done you stick it on a microbalance in an environment with constant temperature and humidity. The microbalance is highly accurate, measuring the weight down to +/- 0.1 μg. That’s 0.0000001 grammes (or around 0.000000004 ounces if you prefer). After two to four days you have stable weight gain and from the measurements you can extrapolate how long it would take for the sample gain its lost weight back.

That might sound simple, but then a lot of the really clever ideas are.

But what about errors?

One of the problems with the write up and press release is that dates were quoted without errors. This makes scientists wonder because errors are part of life when it comes to recording data. If you don’t know how big your errors are then it’s hard to tell how accurate you are. The paper has errors on the dates but these are quoted to one standard deviation. The example they give is a brick from the King Charles II building. This was a built between 1664–1669 and Wren later remodelled it in the 1690s. The date they got was 1691, so that would seem to confirm that the brick came from the remodelling phase. However the standard deviation is +/- 22 years. That means that there’s around a 66% probability that the correct date falls between 1669 and 1713. If you want a 95% certainty, which is what archaeologist should be doing with radiocarbon dates, then you have to go to two standard deviations. There’s a 95% probability that the brick dates between 1647 and 1735. That looks less impressive, but it’s not all bad news. The error comes simply from the number of samples, five in this case, and the resulting average. If you need more accuracy then it should be possible with more samples. Nonetheless this is something to watch for. Any archaeologist who sees a chronology they’ve built their career on overturned is quite reasonably going to ask about the errors.

The other problem is that the dating requires you to know the mean (average) temperature for the ceramic since it was fired. Current research suggests that this could have quite an effect on dates. I don’t think you’re going to be able to get pots fired yesterday with apparent dates of centuries. To get that kind of effect the 2003 paper says you’d need to stick your pottery in an autoclave set to 186ºC for a few hours. But over long periods of time getting the temperature wrong might well put your temperature out by quite a bit. So far the exact effect is unknown, but they note that regional weather records seem to be producing plausible dates.

The temperature problem also works in reverse. Medieval bricks were giving a date of around 70 years. That’s because that’s when the Germans bombed the house they were in, effectively re-firing the bricks and resetting the internal clock. It’s not unusual for houses to have a marked destruction layer when you excavate. In fact ash layers can provide useful boundaries in strata. This could mean that dates from pottery in such buildings will be from the time of destruction, not the time of use. For example I’d expect all ceramics from Herculaneum to appear to date from AD 79.

Isn’t burying a tile at the bottom of the sea going to be a problem?

The first objection that I thought of was for marine archaeology. If this method is all about reacting with water, isn’t going to be a problem if you’re trying to date a tile which has sat at the bottom of the sea for a couple of thousand years? Bizarrely being immersed in the sea might give more accurate results. It’s a matter of the scale at which the reaction works.

Imagine looking at a tile through a microscope. You’ll see all sorts of folds, bumps and even pores leading further into the tile. It’s clear these will be able to fill up with water if you dunk the tile. Wilson et al have been thinking about this and you’d be right. Water would get in and react. Yet what they’ve found that even in the air this effect only takes a few hours to happen. After that all the easily accessible reaction sites have reacted and the effect is over. The rehydroxilation they are measure happens at a much smaller scale, the nanoscale.

If you could look at clay so the extent that you could examine molecules, you’d see it was a tangle of lattice structures and sheets all packed together and twisted around each other. This is where the hydroxylation Wilson et al are measuring occurs and at this scale excess water on the outside of the tile is irrelevant. It’s the temperature of the structure that alters how quickly hydroxyl radicals are incorporated into the clay.

What I am wondering, and a marine scientist could tell me I’m very wrong, is that on the seabed ceramics are going to be protected from variations in temperature to a greater extent than on land due to thermal lag. If this is the case, and we can know that water at a depth X will have a mean temperature Y, then the sea water may actually provide a more stable environment for the reaction and so it could be modelled more accurately.

Bonus features

This technique could be used on all pottery excavated from now on, but I doubt it will. If you’re digging a site which is occupied for centuries, then knowing that this bit of pot was fired on this date won’t add a lot to your knowledge about the site as a whole. You’d be better off spending your budget on something else. Still, for certain key contexts, this technique offers a way to quickly date key pot sherds. The great advantage this technique has is that it seems hard to contaminate. If you know where a sherd came from you don’t have to have it dated immediately. If you’re writing up a report months later you could still send the sherd to be dated if you had records. Curating organic material without contamination on the off-chance you’d want a radiocarbon date would be much more expensive.

It also allows you to date any other pottery which has already been dated. So when did the Minoan civilisation fall? Rehydroxylation dating of pots dug up by Sir Arthur Evans could be dated today to give answers. Suddenly all this material that museums have been holding onto, in case some new technique is invented, is a lot more useful – because this is exactly that kind of technique. The next obvious hurdle is that not every curator is going to be happy with requests like: “Can we smash a bit of this pot please?” to get a sample for dating.

As far as dating goes it might effectively have no limit. Obviously the reaction gets progressively slower as rehydroxylation proceeds, but it seems ceramics have a large capacity to absorb water. Wilson et al cite research that suggests ceramics could add 1% to 2% of their mass through reactions with water, which would mean the method could date material as old as ten thousand years. As far as we know, that’s when most ceramics start being used. There’s some palaeolithic art which may be older, but the chances of being allowed to drill five grammes out of that is so small that it’s not worth bothering about.

So what will it turn upside down?

When radiocarbon dating was adopted it had a dramatic effect on dating. The Neolithic was moved forward and back by a thousand years or more as people discovered that carbon dates needed to be calibrated. The dates are more settled now as there’s a few ways of independently dating material. Dendrochronology has been particularly helpful, so I’d be surprised if rehydroxylation dating suddenly proves the Neolithic is a thousand years older than everyone thought. What it could do is upset some pottery sequences that have gone unquestioned and unexamined for a few decades. It’s been so long since some typological studies that no-one has gone back to check how sound some assumptions are. I’ve a hunch that it could shift dates for Geometric and Archaic pottery in the Mediterranean and upset some sequences there. There are hints that the Dark Ages really aren’t as economically quiet as some historians like to think. I’m sure I can come with plenty of wild speculation, but there is one obvious place where I think rehydroxylation dating could have an immediate effect.

I mentioned quite a while ago that there’s a major debate going on about the settlement of Easter Island. It rests on how reliable you think some radiocarbon dates from coral are. They date from around AD 800, which is when Fenley and think Easter Island was settled. Hunt and Lipo in contrast think that the island was settled in AD 1200. Looking at what is being found elsewhere in the Pacific, and at the problems in radiocarbon dating marine materials I think that Hunt and Lipo are probably right. What rehydroxylation dating offers is the chance to bring new data to the argument. The process would be simple. Get pottery which everyone agrees dates from the earliest archaeological contexts and date it. The error bars are currently wide, but with the disagreement being around 400 years it should be good enough for now. If the pottery can be dated to before or after AD 1000 then we have a winner, with the exact details to be pinned down by further study.

That would be an excellent introduction for this new technique and would certainly make a lot of archaeologists pay very close attention. Well 10/10 for wild speculation. There’s a very good reason why you cannot date Easter Island pottery in the comments below.

The newly published paper is:
Wilson, M.A. Carter, M.A. Hall, C. Hoff. W.D. Ince, C. Savage, S.D. McKay, B. and Betts, I.M. 2009 ‘Dating fired-clay ceramics using long-term power law rehydroxylation kinetics’, Proceedings of the Royal Society A. doi:10.1098/rspa.2009.0117

At the time of me publishing this, the doi is not working. There’s nothing I can do about that, but I you can download the paper from the Royal Society’s own website. The link worked for me, but I don’t know if this is supposed to be a secret page.

Thanks to Moira Wilson at the University of Manchester’s School of Mechanical, Aerospace and Civil Engineering for answering questions that I emailed and correcting an embarrassing mistake I made on the (time)¼ law and to Alex Mack at the University of Leicester’s Centre for Interdisciplinary Science for tracking down the Royal Society link.