Science applied to archaeology and cultural heritage is a thing. It has been happening for decades. Scientific analysis of materials can provide much needed information about materials, trade, manufacture, provenance, foodstuffs, populations, individuals. With today’s kit we can make analyses on tiny samples, or even acquire semi-quantitative analyses without the need for the destruction required to remove a sample. We can identify the components in building materials and pigments, the type of honey a vessel once held, the isotopic signature of metals and therefore their provenance, the isotopic signatures of bone material can tell us were a person lived, where they migrated, where they came from. Science can revolutionise established archaeological chronologies. Amazing information. So why is so much science applied to archaeology and cultural heritage so bad?
I have been to several conferences this year where there has been a cross- and multi-disciplinary approach to materials in cultural heritage. This if course should be a GOOD THING. However, myself and colleagues have been commenting that with the proliferation of analytical techniques and access to them, the science is actually getting worse, and this is not good. We feel that the study of science in cultural heritage is not moving forward. I also see this in papers I review. Scientists are collaborating with archeologists and art historians, but it seems that they are not communicating well and there is little effort on either side to learn about each other’s discipline and what the questions the ‘science’ aims to answer. Inappropriate analytical techniques are used and poor data are produced and these data are, again inappropriately, under- or over-interpreted. The discussions at conferences and recent press activity on the use of a synchrotron to identify the presence of calcium on a Greek vase have spurred me to write this. I admit I have only read the press releases on this latter research, and I presume that the actual publication will provide much more depth to these analysis.
I’m a geologist. My PhD used geothermochronological techniques to look at landscape evolution on a continental scale. I used a radiometric dating technique called fission track analysis which used on the mineral apatite. It was known that the chemistry of apatite, with substitutions between chlorine, fluorine and the hydroxyl group, affected the lengths of the fission tracks and therefore the thermochronometric data obtained. I looked at the subtle changes of chemistry in apatites using Fourier Transform Infra-Red spectroscopy (FTIR). Although I went on to lecture in ‘orthogeology’, much of my research has been associated with the application of geological techniques to cultural heritage. I am so glad – and lucky – to have learned how to use petrology/petrography, geochronology and spectroscopy from first principles, rather than to have stumbled across these techniques as ‘black box’ methods to reveal more about the material I am working on. I understand what these techniques actually tell us about the materials they are applied to. They are measuring bonds and their configurations on a molecular scale, the number of undecayed v. decayed isotopes (or their proxies), elemental weights or excitation of electron energy levels. These data can then be interpreted to give us information such as a radiometric age, they may be diagnostic of the identification of a mineral (or analogous compound) or the nature and origin of an organic compound, they can elucidate the presence or absence of a certain element or the precise chemical compound present in a sample.
The important word here is interpreted. Yep, that’s right, some guy sits down and writes a computer programme to interpret the results that a machine turns out. This is what the software that comes with your new black box analytical machine is. It is NOT the machine telling you the answers. So if you click on the sample image in your back-scattered electron image of your sample (yes, that’s right it is not a photo) and your spectra tells you that a peak is assigned to silicon, this is a programmers interpretation of what fits that peak. Now of course the vast majority of these analyses will be correct and completely reliable, but they should still be used with caution. I have always found pertinent this quote from the 1990 film The Hunt for Red October, in which a submarine sonar operator Jones (played by Courtney B. Vance) discovers that his interpretation software is not telling him the truth. He has detected the faint but rhythmic sounds of another submarine which is using a new propulsion mechanism and is trying to explain this to his commanding office Bart Mancuso (Scott Glen) …
Jones: When I asked the computer to identify it, what I got was ‘magma displacement’. You see, sir, SAPS software was originally written to look for seismic events. And when it gets confused, it kind of ‘runs home to mama’.
Mancuso: I’m not following you, Jonesy.
Jones: Sorry, sir. Listen to it at times speed. [he plays a tape in which a rhythmic noise is heard] Now that’s gotta be man made, Captain.
Mancuso: Have I got this straight, Jonesy? A forty million dollar computer tells you you’re chasing an earthquake, but you don’t believe it? And you come up with this on your own?
Jones: Yes, sir.
Mancuso: Including all the navigational math?
Jones: Sir, I-I’ve got-
Mancuso: Relax, Jonesy, you sold me!
This is so true. That peak fitting software on your FTIR, WDS, EDS, EMP, XRF, whatever was probably not written with reference to archaeological materials. It was written to identify pure compounds used in pharmacy or precise mineral compositions.
So the lesson to learn here is if you get something unexpected, like a massive peak assigned to tellurium, you should probably question your results, not just accept it as something really unusual and therefore cool. Any scientists watching your presentation or reading your paper (or indeed reviewing it) will immediately see through this. Here is an example from the field of petrology. What you need to know before reading this is that basic mineral identification in rocks is carried out by (experienced) analysts using optical polarising light microscopy (PLM). So for example, a basalt may include a mineral from the pyroxene group where chemistry ranges between iron, magnesium and calcium and variable silica amounts. This group can be conveniently subdivided into two main groups orthopyroxenes and clinopyroxenes which have clear association with different geochemical environments. The ‘ortho’ and ‘clino’ bits refer to differences in crystal structure and these two groups are most easily distinguished using PLM. It would take about three seconds to distinguish the two pyroxenes and we generally just refer to them as orthopyroxene (opx) and clinopyroxene (cpx). If you really want to, you can further subdivide the pyroxenes into named types, say for cpx, the main ones are are augite, hedenbergite and diopside. However most of the time we geologists don’t use this classification unless there is a reason for doing so, i.e. we want to answer specific questions about zoning within a single crystal and how that relates to say, fluctuating melt chemistries (and we need a reason for deciding that fluctuating melt chemistries are important). We just call it clinopyroxene. Therefore when I see a presentation that shows someone has identified diopside in a pot sherd I know they haven’t looked properly at the material and a computer programme written for igneous petrologists has told them it is diopside and they didn’t question it. It is also important to note that the name ‘diopside’ does not denote a specific phase chemistry that will direct you to a particular source/provenance. It doesn’t and it won’t. And it isn’t ‘more scientific’.
Clinopyroxene. I really DON’T CARE whether it is diopside, augite or hedenbergite.
On the subject of mineral names, the misuse that infuriates me most is the use of the mineral name cuprorivaite to describe the synthetic pigment Egyptian Blue. If you Google cuprorivaite, you will get far more references to Egyptian painting than to minerals and this demonstrates the scale of this problem. Cuprorivaite is a rare, naturally occurring mineral. When it occurs it is as a micromineral (very small crystals, < 1 mm) and often disseminated. It occurs nowhere in masses worth of economic extraction, even as a cottage industry. Therefore it has never been used as a pigment and no one has ever detected cuprorivaite on archaeological paintings. However, you would not believe this from the literature. What researchers have found is an analogous synthetic compound called Egyptian blue. It’s a calcium copper silicate if you want to sound scientific. Why is this terminology important? In the world of pigment analysis, there are many phases which can and have been used as pigments derived from natural minerals and analogous synthetic compounds and it is important in answering archaeological and art historical questions to be able to distinguish these forms. Examples are the red pigment mercury sulphide which when natural is cinnabar and when synthetic is vermillion, or (natural) azurite and (synthetic) verditer for the blue copper carbonate hydrates. It can be important to know the differences between these minerals to detect fakes and assess knowledge of technologies or mineral provenances. As a consequence it is essential to have a terminology that clearly differentiates between natural and synthetic pigments.
A photomicrograph of the mineral cuprorivaite.
I think a major problem here is that a lot of the time, scientists and cultural heritage people don’t really get each other. They don’t know enough about each others subjects. Sure, they watch the TV programmes and remembered being into the Ancient Egyptians at school, or they maybe had a toy chemistry set. But now we are all grown-up academics, we are all set in our ways and stuck in the silos that the university academic departments give us (and this structure is a problem). A chemist is probably not likely to meet an art historian at their institution unless they sit together on a college committee – and many academics will avoid these committees like the plague. So we don’t talk and learn about other disciplines, and don’t get me wrong, scientists are just as bad about this as anyone. Many can be arrogant, thinking that they can answer the ‘simple’ questions archaeologists ask and they think its cool to have something they have only seen before in a museum or on the TV in their lab. Something to tell their colleagues and hey, it might just tick some of those ‘impact’ boxes. So an art historian comes along with a Greek vase and the scientist thinks ‘OK I can analyse this for you using my synchrotron. This will cost hours of beam time and thousands of dollars, but we may get a paper from it that will enable me to patronise humanities people with my superior knowledge and show that I can also do novelty science. I have seen Greek vases in museums, I am surprised that no one has made these analyses before because it is so easy to do’. The art historian thinks ‘Brilliant, this will tell us something that all those other techniques won’t tell us. I don’t really understand those techniques, but surely this synchrotron thing is so big and so expensive that it must provide better, richer, more accurate and more precise results than anything else. This will make me look amazing when I present this work! So few people in my position have access to this sort of kit. Just using this technique makes this study novel and innovative. And the scientist guy says it will be easy. He seems to know what he’s doing.’. Maybe, just maybe, one of the parties will search the literature, but because scientific literature database on the whole doesn’t search books and memoirs and vice versa, they don’t find out that this is something that has already been done, using simple analytical techniques which have given better and more informative results.
And this happens within science disciplines too … this (modified) cartoon is on our geochron lab wall. ‘We’ are the geochronologists …
Despite what you may interpret from the above, a geologist or geochemist would never use complex, expensive kit to do routine mineral analyses, so why would you do this to analyse the pigments on a wall painting or on a Greek vase? Many simple techniques such as microscopy and wet chemistry give reliable, accurate and precise results and can’t be improved on. The press releases surrounding the analysis of pigments on an Attic lekythos carried out on a synchrotron only told us that the white pigment present ‘contained calcium’. It stopped short of making the logical (i.e. totally f**cking obvious) observation that this was a form of calcite, a polymorph of calcium carbonate. If I could have taken a tiny sample from the vase (so small that it could not be noticed by the naked eye) I could have told the museum/art history guys that not only was it calcite, but would have been able to differentiate chalk, eggshell, seashell (OK the aragonite polymorph in this case), coral or any other form of calcite derived from a limestone. If sampling was not possible, I could have identified the presence of calcite/aragonite (very unlikely that it would be aragonite though) using a UV light torch and a pXRF in less than a minute. I would have probably charged £50 max for experimental costs. To present a paper on pigment analysis and conclude that you have a Ca-rich pigment or an Fe-rich pigment is simply not good enough and is not moving the subject on.
Mind you, on the subject of portable X-Ray fluorescence machines, these will only give semi-quantitative results and users should be aware of this, but they are good for identifying major elements when sampling is not possible. However pXRF analyses should not be used in isolation. These machines were designed for quick, rough, field checks to look at say, arsenic pollution in ground water. The manufacturers are horrified at how their data has been used to provide ‘confirmed scientific analyses’ in cultural heritage.
My examples here are mainly to do with geological and analogous materials but I am sure that many scientists can quote examples from metallurgy and biosciences where glaring misuse of equipment and terminology are used. How can this be improved? We all have to learn more from each other. Most importantly cultural heritage people need to know what questions they want to ask of their objects and materials. Is it as simple as ‘what is it?’ (a very valid question) or do they want to know more; where does it come from? How was it made? If there are more complex questions to be answered then find a scientist that can work with you to help find the best analytical method to get the answers you need.
My top tip is that scientists and cultural heritage partners work together in a truly collaborative manner, so that the scientists are not just used as technicians, they are an integral part of the research project. I have always worked in close partnership with colleagues and have made time to learn about the materials I am analysing and the contexts of the artefacts or building from which they are derived. I know about comperanda, I read the literature, I know what the expected range of materials are. I double check my results when I find something outside that. If its true I design further experiments or do field work to answer these new questions. I don’t just hand over the data and walk away. And I always, always use the most appropriate technique to perform scientific analysis. As a petrologist I am experienced with polarised light microscopy so that is my ‘go to’ technique. It is great for ceramics, mortars, pigments and of course, stone. I do accept that this takes learning and experience, but go on a course! Learn it! I do a lot of work on pigments so I use FTIR and Raman spectroscopy too. If I have questions about organic binders then I would turn to gas chromatography mass spectrometry, but as this is costly and wouldn’t go there unless it was really important. I use old-skool spot tests and wet chemistry for quick analyses of phases to detect things like lead or phosphate (there are loads of books on how to do this, and you can do it in you kitchen). I use pXRF and SEM/EDS, again for major elements only, but am cautious on how interpret results, knowing these are not quantitative techniques. I occasionally use X-Ray diffraction (XRD) to look at crystallinity in some minerals (mainly hematite and other iron oxides and iron oxide hydroxides in ochres). These techniques answer all the questions I can think of. I used a synchrotron once to try and design some experiments to further the understanding of the blackening of cinnabar/vermillion paints. It didn’t work.
I used a synchrotron to analyse a pigment and all I got was this crappy picture.