9.3 Chronology

Scientific dating

The domestic character of the Iron Age generates considerable quantities of good dating material, even from antiquarian excavations. Taken with Bayesian modelling, this has the potential to revolutionise the understanding of both important sites and wider sequences, and how to take advantage of this deserves further consideration.

Radiocarbon dating hinges on adequate sampling strategies and the need for long sequences of single-entity dates (Haselgrove et al. 2001, 5) where possible. There are several exemplars of good practice (e.g. Lelong and MacGregor 2008; Hamilton and Haselgrove 2009). Problems with the calibration curve for periods of the Iron Age can be offset through good sampling strategies and the application of statistical techniques.

There would be great value in obtaining dates from securely-stratified material from both old and more recent excavations; the stratigraphic control from post-War but pre-AMS days should allow classic sequences to be tightened up considerably by obtaining fresh series of dates. This is a cost-effective approach to maximising value from existing archives, and would be best targeted to regional sequences. Notable possibilities (and outstanding problems identified in other sections) are the Western Isles drystone architecture sequence (where the Rocket Range wheelhouse site, Dun Mor Vaul, and sites such as Berie and Bharabhat on Lewis have the combination of good excavations, good organic survival and a limited or uncertain range of dates at present), Howe on Orkney, and perhaps the Border hillforts sampled in the 1940s and 1950s (see theme 2).

The potential of other techniques must also be remembered. Recent work sampling wood from unexcavated crannogs on Loch Tay and Wigtownshire highlights the potential of using dendrochronology to get outline dates for a large number of sites to obtain broad patterns. A range of other techniques have potential for the Scottish Iron Age, such as archaeomagnetism, thermoluminescence (TL) and optically-stiumlated luminescence (OSL) (see Dockrill et al. 2006 for an integrated approach to dating with a range of techniques at Old Scatness) offer a suite of approaches to deal with Iron Age material. Their application requires consideration of the type of material available and the levels of precision required. Pottery, for example, is present but not abundant in many parts of Scotland, so there is less potential than other parts of UK for getting multiple determinations through TL from single sites to refine. Scottish Iron Age archaeologists are also perhaps less trusting of alternative radiometric dating techniques because of the conflicting results produced by studies in the 1980s on vitrified forts (e.g. Sanderson et al. 1988 ; Strickertsson et al. 1988 etc), though more recent successes may be changing attitudes (e.g. Anthony 1999).

A problem for the detailed reconstructions of Iron Age life advocated in this report is that it is difficult to narrow down the occupation of sites to the extent that contemporaneity can be confirmed. While many sites may have been occupied over a period of decades, if not hundreds of years, Halliday (2007) has suggested that the Butser experiments - showing that a large wooden roundhouse can be maintained successfully for several decades - have misled archaeologists into expectations of longevity for later prehistoric sites in general. On the basis of evidence such as a lack of multi-phase hearths, he suggests that the upland hut-circle ‘settlements’ of Scotland actually represent the remains of far fewer families than is normally supposed, with families abandoning each house after a period of occupation of five years or so and building (or re-inhabiting) another house nearby (Halliday forthcoming; cf. Roymans and Gerritsen 2002) (see theme 5 where evidence of longer duration sites is also discussed). A close chronology allowing assessments of contemporaneity, is key to reconstructions of society and landscape use, and Bayesian analysis of securely stratified, short-lived radiocarbon dates now offers a way forward, with probabilistic assessment of competing models (see Whittle et al. 2011 for a Neolithic case study). This Neolithic example has shown that phenomena previously thought to have been spread over several years were actually restricted to a handful of generations, essentially pulses of activity, which has major impacts on any understanding of the period. Exploring this requires more and better dates, but would be a key aspiration of any redating programme.

Targeted sampling of monument types has produced valuable results in the case of Strathdon hillforts (Cook 2010) and hut circles in Skye (Wildgoose, pers comm.). As discussed in theme 6.3, it provides a means for rapid and cost-effective provision of a basic dating sequence, although caveats over its small-scale nature must always be remembered. Such programmes could provide valuable first-stage chronologies in many areas; their value is in the broad picture, not the detail of individual sites.

Robust systems of systematic environmental sampling should provide the material for a C14-defined chronology of most excavated sites. These should be analysed within a Bayesian framework, and compared to other sites to build local and regional models for further testing.

Whenever samples suitable for dendrochronology are encountered, the opportunity should be seized.

There is tremendous potential in exploiting existing excavated archives for redating programmes, analysed with Bayesian methods. Carefully-constructed research programmes should allow generation-scale study of Iron Age societies in selected sites and regions.

Keyhole excavation provides a more partial view of site chronology, but has value to obtain a first-stage pass at settlement sequences in an area.

Artefactual dating

Chronological patterns in later prehistoric material culture are poorly understood. Lowland pottery has remained stubbornly resistant to the diagnosis of chronological and cultural distinction. Atlantic Scotland shows more ceramic variation, with a pottery sequence proposed for the Western Isles (Campbell 2002), and progress with the analysis of Northern Scottish and Northern Isles material, though much work remains, and there are signs of significant local variability (MacKie pers comm.). With other artefact types, knowledge of chronological change is still poorly developed – well-dated contexts are needed to help the dating of finds rather than vice versa. The problem is well-exemplified by the dating of long-handled bone and antler combs, an outwardly diagnostic type, but one which remains in production and use in Atlantic Scotland much later than elsewhere, and 'Pictish' painted pebbles, now shown to be an Atlantic Middle Iron Age type (Sharples 2003, 154; Dockrill, pers comm; Goldberg & Hunter forthcoming). Needham et al. 1997 provide a good example of how such an approach radically improved the dating of an artefact group (in this case Bronze Age metalwork).
 
The dating of Caulfield’s (1978) “quern replacement horizon” from saddle to rotary querns in the Western and Northern Isles is contentious, though it appears to be a phenomenon observable across the country (MacKie 1987, 7). Caulfield himself did not suggest a date (although he argued that it was a more or less immediate changeover) but MacKie (1987, 7-9) posited a date in the first or second century BC. This was subsequently taken up by Armit (1991, 192), who argued for a date of c. 200BC. Yet in Southern Britain rotary querns were in use by the 5th century BC, and there are hints of similar early examples in Scotland, th The direct replacement of saddle qough the evidence is not watertight (see theme 4.5). The direct replacement of saddle querns by rotary querns is harder to prove as the former remained useful for processing materials other than grain long after the introduction of the rotary quern (Armit 1991, 192). Advances in considering the date of introduction of the rotary quern are complicated because almost all querns come from secondary contexts, and often after long lives, giving only a terminus ante quem, but a review of evidence emerging from recent excavations would be well worthwhile.

Although earlier scholars considered indigenous and imported metalwork as a potentially valuable dating mechanism, there have simply not been enough finds from secure, radiometrically dated contexts in Scotland to confirm the authenticity of the suggested date-ranges. However, there is no good reason to think that extended, , ‘south-north’ time-lags occurred. Modern analyses have tended to err on the side of caution in consideration of such goods (e.g. Hunter 1998a, 346; 1998b, 393); direct dating of objects and contexts associated with them is helpful, but more data are needed as the sequence is not yet clarified (Garrow et al. 2009). More helpful are metal alloys, and especially the presence of zinc; this seems to be a reliable indicator of re-melted Roman alloys, thus providing a terminus post quem (Dungworth 1996).

Traditionally, Roman finds were used for dating Iron Age sites, though obviously they can only provide dating evidence for the period of Roman contact and later, or more often act only as termini ante or post quos. This has sometimes been undertaken assuming a simple model of the use of Roman material culture, with rapid intake and rapid discard. It is now clear that this needs to be assessed by independent dating, not assumed and that the possibility that Roman artefacts continued in use long after being acquired very often needs to be borne in mind when considering their chronological implications (e.g. Alcock 1963; Alcock & Alcock 1990, 115-6; Warner 1976, 285-8; cf. Hunter 2007a, 11, 91).

The typologies and dating structures of glass jewellery both predate radiocarbon, and as a result are heavily tied to the occurrence of Roman finds – the danger is that this creates a misleadingly short and late chronology. Although there is a basic typology for the region, the conventional typological dating of native glass beads by Guido (1978) was based on a diffusionist model. Blue glass beads, in particular, seem to have had a long currency, stretching from the Iron Age to the early Historic period. Glass bangles (Kilbride-Jones 1938; Stevenson 1956, 1976) have potential as a dating tool, though the extent to which Roman vessels were reworked as raw material is unclear, and obviously has implications for dating, with earlier dates possible if indigenous manufacture (perhaps from imported ingots) is accepted.

Revisiting typological schemes offers considerable potential – there are changes in material culture through time, but current understanding is very poor, as radiocarbon is rarely targeted to this question. Yet direct dating of objects (e.g. organic objects and residues on diagnostic pot) has the potential to revolutionise this situation, and in the process greatly enhance the dating potential of artefacts. The targeting of deposits producing interesting finds for dating should be an important aspect of post-excavation work.