Time-Off in Roman Iron Age Skälby
1 December, 2014
This week On the Reading Rest I have three reports from the excavations at Skälby, the last of the three Västerås settlements to merit a Bayesian chronological analysis. All three reports have an English summary. The two first reports concerns the Skälby village, the last one deals with excavations in the outskirts west of the village. I read these reports mostly to understand how the 14C -dates mirror the chronology of the settlement.
Onsten-Molander, Anna. 2008. A. Onsten-Molander (red) Skälby. Bilden av byn växer fram. Fortsatta undersökningar av boplatslämningar från äldre järnålder. Med bidrag av Ylva Bäckström, Tony Engström, Susanna Eklund, Helena Hulth & Ann Lindkvist. SAU Rapport 14. Uppsala 2008. http://www.sau.se/filarkiv/rapporter/saurapport_2008_14.pdf Acronym OnMo.
Korpås, Ola & Wikborg, Jonas. 2102. Skälby. Bebyggelselämningar från järnåldern. Med bidrag av Elisabet Pettersson (Skälby. Settlement remains from the Iron Age (with contributions from Elisabet Petterson). SAU rapport 2012:13. http://www.sau.se/2012_13_skalby_a661_s198.html Acronym KoWi.
There are c. 120 14C-dates related to the greater Skälby settlement area collected between 1992 and 2009. They describe a settlement period with varying intensity. The most common 14C-dates belong to the PRIA. In the LPR- and ERIA the settlement consisted of a village just south of a small impediment where the remains of an IA cemetery are still visible. West of this EIA village there are some small settlement areas on the fringe of a small forest growing on moraine partly covering a very small mountain. It has been suggested that there was a small prehistoric ring fort on the mountain (KoWi:10). Village and outskirts shared an agricultural area between them with small heavily fertilized fields and water meadows (Asp:71f.). The arable land is often wet and dominated by a clayey raised seabed. To the north these fields meet slightly higher woodlands on moraine. The arable land stretches out towards the south and the small stream between Skälby and West Hacksta.
Historical and modern Skälby is situated just east of the Early Iron Age village. Today, the area around the modern village is surrounded by the suburb Skälby, but it stands to reason that the LIA settlement lies under the modern village or north of this village next to the remains of the cemetery. Since the latter alternative creates a connection between excavated settlement remains and the cemetery, this might well be where the LRIA village was situated.
The distribution of the 14C-dates has a strong emphasis on dates belonging to the period bp 2150–1950, i.e. a c. 200 year phase commencing in the early 2nd c. BCE and coming to an end in the 1st c. CE. All periods from the Late Bronze Age (LBA) to the Late Carolingian Iron Age (LCIA) are represented in the area, but this cannot hide the fact that from the 4th century and onwards dates become scanty. The emphasis in the village on the LPRIA is enhanced if we take the dates from the peripheral settlement area into consideration. No less than forty tests, a third of the total number, are dated bp2100–2000 (2nd–1st c. BCE).
Slightly later, between bp 1960 and 1880 there is a possible gap in the 14C-distribution and it is worth testing whether this gap in the village settlement actually existed. The models used to test this will be returned by the BCal – an on-line Bayesian radiocarbon calibration tool (1). The solitary 14C-date (bp 1920) in the middle of the gap actually doesn’t represent the settlement inasmuch as it belongs to a peripheral well. It testifies to the fact that the area, whether settled or not, was continuously used, albeit with varying intensity.
The Skälby 14C-dates from the LPR- and RIA are not easy to model. They are many and since most of them were measured in early 1990s several average values are afflicted with standard deviations twice or thrice as large as the present standard. This makes modelling a possible gap or hiatus difficult. If we look at the end of the long, mainly LPRIA, settlement period and the beginning of the short RIA settlement (cf. the green circle in the above illustration), then the gap is hard to prove because the limits of a short period are more difficult to define than the borders of a long one. If we check the gap only in relation to the nearest 9 tests on each side of it, the gap is not like to have existed. Moreover, if we try to compare the 45 often clustering tests of the whole LPRIA period to the 12 tests representing the short RIA period, BCal will fail to calibrate the sample. However, if we delete every second test from the sample of the long period, then modelling becomes possible because the length and character of the period is retained while in the process the number and complexity of the tests have become manageable. When BCal is asked to calculate the gap between the event that signifies the end of the long period and the event that signifies the beginning of the short period, then the tool returns the interval 2 to 181 years if the probability is set at 95% and 52 to 138 if it is set at 68%. The fifty-fifty length is 65 to 120 years. An interval of c. 65 years is thus not unlikely although the gap may have been longer as well as shorter.
By checking the possibility that the long Phase A had come to an end and the short Phase B commenced, in relation to given calendar years, we may illustrate when the gap was likely to have been a fact. It would seem that at least in the later years of the 1st century CE there was a gap in the Skälby settlement. A comparison between the posteriority distributions of the end of the long settlement period, Phase A, and the beginning of the short, Phase B, give the same impression. This, obviously, is a result that ought to be reviewed in relation to the spatial contexts of the dates.
If we look at this distribution of the 14C-dates sorted in 50-year long 14C-year periods, the development of the village stands out. The same is true of peripheral wells and a tendency for the settlement to form phases which probably has to do with digging or repairing wells close to the farm houses and the rebuilding of farm the houses themselves. Generally speaking Skälby is not a very dynamic settlement since renews itself in phases rather than continuously; but it accords with a situation in which a settlement is abandoned in the late 1st century CE and eventually reestablished as a number of well-separated homesteads. A settlement that can be switched off and on—abandoned and reestablished – doesn’t stand out as autonomous. In the second century CE, not surprisingly, the distribution of the farmsteads is very different from that of the early LPRIA settlement, which had central and peripheral parts.The short RIA phase is a transition to a long period (10 14C-dates bp 1790-1600) of limited usage of the settlement site. An intuitive analysis of the dates makes it likely that among four farmsteads, crofts or cottages, only one exists in the settlement area c. 1650bp when the settlement was finally abandoned. For more than a century, this farm or croft had been situated in the northern end of the area next to cemetery. With a series of main houses c. 15 m long, this farm seems to have been established c. 1775bp and abandoned c. 1650bp, that is, c. 400 CE. During the settlement period bp 1790-1650 there seems never to have been more than two households in the area. Owing to the calibration curve the oldest dates in the period may well be more or less contemporary with the latest dates in Phase B in the northwestern corner of the excavation. This indicates that in the end phase (bp 1790-1600) there was one, two and eventually one continuously settled abodes in the northernmost part of the settlement area — on slightly higher grounds next to the cemetery. It is possible therefore that the end of the Skälby settlement is actually the end of the outskirts of a village situated north of the excavated area. If so, the village might well have looked like the last phase in West Hacksta, Village E (cf. OtRR 3 Nov 2014).
An Outline of the settlement history in the Gilltuna-Hacksta-Skälby triangle
In order to summarize the development in the settlement area I will interpret the chronological events as typical general events. This need not be the case, but it is a reasonable way of creating a model that may be falsified by new excavations.
In a long-term structural perspective, the settlement area develops from the PRIA and onwards with the isostatic uplift, expanding in terms of settlement units until the RIA, seemingly starting to disappear in the 4th century CE. The development in Gilltuna gives us a glimpse of the LIA organization of a settlement. Most importantly, the densely settled ‘tun’ illustrates the concentration of buildings on a LIA plot. On the tun in Gilltuna the remains of 17 houses covering c. 500 years, could be defined on 5,000 sqm. At Skälby, 39 houses were found on 70,000 sqm during an equally long period. This means that there was one house per 1800 sqm at Skälby. At Gilltuna during a similar time span there was one per 300 sqm. The relative density of houses on the tun was thus six times higher at Gilltuna. To this one must add that it is much more difficult at Gilltuna than at Skälby to link all large postholes to houses. Thus there are probably more unknown buildings at Gilltuna that at Skälby. Structurally, the organization of the settlement in the middle of the first millennium is thus a matter of confining farms to stable regulated narrow plots. The sites chosen to become dense regulated plots, and thus probably a ‘tun’, were used already in the EIA when it seems that most of the suitable settlements sites had already been recognized.
Moving a settlement, was part of the cultural identity of the EIA, the restricted and permanent plot on the other hand was novel and probably introduced in some places already in the LRIA, for instance in Village D162 in Säby, Uppland (cf. OtRR 13 March 2013).
Between the end of the 10th century and the 15th, when historical Gilltuna disappears, there is only one date of interest, an oven dated bp 486±30, i.e. cal CE 1407-50 (with ±2σ). This indicates that the remains of Early Medieval Gilltuna with most of its buildings standing on the ground was ploughed away from the 1400s and onwards. The dates from the excavations of the peripheral sites west of Skälby suggests that in the CIA peripheral activities were to some extent revived. If we summarize the analysis as a matter of structure and chronology, as in the above illustration, we can describe a generalized development with rotating farms in the PRIA. This period of expansion leads to the first small villages in the end of the PRIA. In the ERIA villages are sometimes reorganized and a clear divide between central and peripheral farms, as well as crofts, becomes a reality. In the LRIA some villages become a little more regulated and in the end of the RIA and in the 4-5th century many villages disappear, but not all. Many peripheral settlements, moreover, continue to be inhabited before they finally disappear c. 500 CE. It seems reasonable to suggest that Gilltuna continues to be Gilltuna although it may have changed its name after the settlement hiatus. IA Skälby lives on in historical Skälby and West Hacksta may eventually have become Igelsta.
Settlement concentration and reorganization starts already in the RIA and it is an ongoing process which deprive us of small settlement units. Eventually the settlement contracts to a few densely settled village sites. Since we can see that this village development starts already in the ERIA, we might have expected it to be a gradual process, but the abandonment of a large number of farm units in the late 4th early 5th century might well represent an agricultural crisis speeding up the development. The Cold Decade didn’t stop the development although one might have thought that a period, starting with a drop in the population, would lead to the foundation of new farms when the population began to grow again. If so, these new villages were successful and invisible to constract archaeology, with a few cases such Gilltuna to prove the rule. The Klondyke situation characterizing the PRIA was not repeated in the LIA and not until the CIA do we see signs of expansion.
If we summarize the analysis as a matter of economy and chronology, then we must first acknowledge that in the long run farms become fewer, larger and more stable. Some farms, moreover, become larger than others. During the PRIA, husbandry is important, and fields are small and over-fertilized. When the households become larger it stands to reason that the fields grow too and the divide between fields and grassland more marked. Owing to larger fields and denser settlements, the balance between fields, grass- and woodland becomes stable and more prolific. The nucleated villages will benefit from larger fields and roads from the villages through infields to meadows and woods.
The social implication of the development stands out as social segregation and the transition from a flat to a more pointed social pyramid. The introduction of larger farms, dominating a village, and the ability to prevent peripheral settlements and settlers estabishing themselves outside the villages is a major social achievement that probably reflects the power of those who think the land belongs to them. They, who in the LIA live in pit houses outside the Gill-tun, either permanently or seasonally, are probably less socially important than those who live in halls. The former represent a growing population with no common right to settle and farm a suitable unoccupied land.
Owing to possible subsistence problems, the closing down of autonomous households in the 4th and 5th century, and the possibly famine in The Cold Decade, there are also demographic implications in the changing landscape. Migration from the area and trying one’s hand at external acquisition, as well as power struggle may have been seen as ways of coping with crisis, at the same time limiting the growth of the population. The Cold Decade, although a purely natural phenomenon may thus have hit a population that was badly prepared to resist it when crops failed and grasslands and meadows became low-productive.
(1) The BCal team comprises Caitlin Buck, Geoff Boden, Andrés Christen, Gary James and Fred Sonnenwald. The URL for the service (http://bcal.sheffield.ac.uk). The paper that launched it was Buck C.E., Christen J.A. and James G.N. 1999. BCal: an on-line Bayesian radiocarbon calibration tool. Internet Archaeology, 7. (http://intarch.ac.uk/journal/issue7/buck/).