Mapping the Stonehenge bluestones with mineralogy
Where did the Stonehenge bluestones come from? Scientific advances are allowing us to pinpoint the outcrops that they were quarried from with ever-greater accuracy. Rob Ixer, Richard Bevins, and Duncan Pirrie describe some of the latest thinking.
The quest for the geographical provenance for the stones of Stonehenge is centuries old, and became scientific by the middle of the 19th century – so that, in 1883, Wiltshire antiquarian W Cunnington, could write: ‘Geology and petrology have lately joined in the pursuit of time-hidden truth, and the microscope has been used with some success in this direction, but much remains to be done’.
Almost 150 years later, that statement remains just as cogent. For Stonehenge, the last decade in particular has shown that routine microscopy has had its successes but may have reached as far as it can go in provenancing the monument’s bluestones. It is now clear that additional advances will only happen by applying new mineralogical, petrographical, and geochemical techniques alongside established methodologies.
The provenancing of the Stonehenge bluestones (these are all the non-sarsen stones and include both igneous and sedimentary rocks) used to build the monument has to be founded on their accurate and detailed characterisation. Only then is there hope for tracing them back to their source outcrops. We have been doing this (initially for the igneous bluestones) by combining thin-section petrography and geochemistry for over a decade. Through this method, we have managed to determine and identify the number of different lithologies found at Stonehenge, some very recently, both within the standing stones themselves and also among the many thousands of pieces of buried debris – the so-called ‘debitage’ – found scattered throughout the Stonehenge landscape.
Notable successes include recognising that the rare clay-like mineral stilpnomelane, found in debitage, matched that from rhyolitic rocks near Pont Saeson, on the northern slopes of the Mynydd Preseli in west Wales. This led Mike Parker Pearson and his team to identify the probable Neolithic quarry site at Craig Rhos-y-felin (see CA 311 and 345). However, although this rock-type (now designated Rhyolite Group C) is highly abundant as debitage, it only seems to match one of the buried orthostats (standing stones). Could there have once been more, with other Rhyolite Group C orthostats now totally lost, with only the debitage to suggest their former presence?
Similarly, the presence of the rare phase graphitising carbon in orthostat SH38, and a characteristic feldspar alteration in orthostat SH48, allowed debris from those two standing stones to be collected, counted, and its distribution within the Stonehenge Landscape evaluated; surprisingly, the debitage from both stones is very rare, less than 1 per cent, but nevertheless quite widespread. But not all the lithologies have helpful exotic minerals or textures. These have proved to be more challenging, and as we started our second generation re-examination of the bluestones we realised that more systematic qualitative and quantitative work was going to be needed if we were going to continue to be successful in matching them to their outcrops.
Quantitative modal mineral analysis (finding the absolute and relative amounts of each mineral in a rock) is a powerful tool in highlighting the similarities/dissimilarities between rock samples. However, there is a cost: identifying each grain optically and manually point-counting them is slow, tedious, and error-prone. Clearly a different identification method was needed, and an innovative ‘automated mineralogy’ approach has become pivotal to our re-examination.
WE HAVE THE TECHNOLOGY
Many electron microscopes used in the Archaeology and Earth Science departments of universities and museums will have an energy-dispersive spectrometer, which can be used to determine the chemistry of individual spots on a sample. In recent years, though, a range of advanced automated SEM-EDS systems have become available. These allow individual polished sections to be mapped at a level of detail previously not possible, with very rapid automated acquisition of EDS chemical analyses across a sample. These EDS (Energy Dispersive X-Ray Spectroscopy) analyses are then compared with a library of spectra of known mineral chemistries, and each
analysis-point is assigned to a mineral species. Not only does this enable us to generate huge amounts of data (up to 3.5 million datum points per sample), the mineralogical information is also in context, as the samples analysed are effectively mineralogically mapped. In addition to identifying each mineral grain, it gives quantitative modal mineralogical data, even for fine-grained rocks. This greater accuracy has become instrumental in confirming and also establishing separate, and now tightly defined, bluestone lithologies.
Stonehenge’s non-doleritic volcanic bluestones are fine-grained, and are either quartz-feldspar-rich or have been altered to be clay- or secondary mineral-rich; the spotted dolerites too are altered to fine-grained secondary minerals (especially calcium-aluminium-rich hydrous silicates), although some primary igneous minerals are preserved, especially clinopyroxene and locally calcium-rich plagioclase. It proved difficult to optically differentiate between fine-grained quartz and untwinned feldspar, or between the many minerals formed during the alteration and weathering of the bluestones – hence our need for an alternative technique that accurately and correctly identifies them chemically, rather than by their optical properties.
One of the first ways that we used the automated mineralogy approach was in re-examining the iconic white/pink spots found in the majority of the Stonehenge dolerites (the material of the most-common bluestone orthostat, and second most-common debitage). Almost exactly 100 years ago, it was these spots that led H H Thomas to recognise that the Stonehenge dolerites should be provenanced to the Mynydd Preseli area in west Wales. But the petrography of the spotted bluestones is very similar, both among the Stonehenge orthostats and within the Preseli’s outcrops, so much so that it proved difficult, in thin-section, to be able to link orthostats to specific named outcrops with any certainty. Instead, we found that whole rock geochemistry proved better and more discriminatory, and this approach led us to suggest that the outcrops at Carn Goedog were the origin of most of the spotted bluestone orthostats; subsequently, and once again, a Neolithic quarry site was recognised from there by Mike Parker Pearson and team (see CA 311).
Since the spots vary macroscopically in size and colour between outcrops on the Preseli hills, it might be that a detailed re-examination of them can further refine their origin. In thin-section, these spots (constantly misidentified as feldspar) comprise alteration products with very similar optical properties. But SEM-EDS analysis holds the key to distinguishing them, showing them to be in fact fine-grained mixtures of secondary feldspar and clinozoisite, along with minor epidote. It may be that detailed modal mineral examination of the spots between different dolerites, both at Stonehenge and in the Preseli, can be used to further refine their provenance – something that could help to identify further potential quarry sites for archaeologists to investigate.