In addition to the problematic nature o f small town definition, the data pertinent to the small towns in this study have their own set o f problems. These troubles are associated with the nature o f archaeological data as well as the evolution o f archaeological thought and methods over the century. Together these forces have complicated the analysis and required a
methodology specifically designed to compensate for these problems and illuminate the economic and social change occurring in Britain during the Roman era.
To better understand data pertinent to this study it is important to have an understanding o f the building types and construction techniques used to make them and the traces they leave in the archaeological record. The diversity o f buildings in small towns is relatively limited as compared to the larger towns, the majority being rectilinear domestic/workshop combinations (Burnham 1988, 54; Burnham and Wacher 1990, 17). The choice o f construction techniques was driven by both economic and social considerations and traditions (Burnham 1988, 39-40;
Hingley 1989, 31; Burnham and Wacher 1990, 17; Perring 2002, 80), though it should be kept in mind that the choice o f construction techniques may not have always rested with the owner since the literature and legal codes indicate that were significant planning controls (Perring 2002, 80). De la Bedoyere (1991, 21) postulates that the majority o f buildings were constructed o f timber, easily overlooked in the archaeological record. However, it does become clear that masonry was increasingly adopted in second-century (Buckland 1988; Blagg 1990, 48; Perring 2002, 106).
Circular buildings were an Iron Age tradition of vernacular domestic structures that has continuity into the second-century in the small towns and longer in the more rural countryside, continuing in use beyond Hadrian’s Wall and also in Wales (Burnham 1988, 38; Hingley 1989, 31-35; Burnham and Wacher 1990, 17). The buildings are constructed by placing vertical
stakes or posts directly into the soil w hich is packed around it for support. W ith a few notable exceptions, m ost circular buildings have an absence o f R om anized features (Burnham 1988, 38;
Burnham and W acher 1990, 17).
‘ T S ’
( : )i T t
jn
G s~- J • • *“■ v.-*
f e - e - e -
/ S I ' /
h l
• rr r
o s y e t r e s
Fig. 2.1: Types o f B uildings in R om ano-B ritish Sm all Towns (from Burnham and W acher 1990, 16)
A to C: C ircular B uildings
D and E: Stone or stone founded rectilineral buildings F: an exam ple o f building know n m ostly from the flooring re m ains
G to 1: Earthfast construction
R ectilinear buildings are the m ost com m on building in small towns (Burnham 1988, 38).
D om estic structures becom e larger over the R om an era and usually had spacious plots
associated with them (Burnham and W acher 1990, 18). A nother type o f rectilinear building, the
“strip” building, is also very com m on. Strip buildings w ere placed w ith their end on the street, often with an open face. The spacing betw een buildings was m inim al indicating that street frontage m ay have been at a prem ium . The general interpretation is that these buildings w ere w orkshops or stores with dom estic quarters for its ow ners (Burnham and W acher 1990, 18). A
third type o f rectilinear building, generally termed an aisled building, had twin rows o f internal roof supports (Hingley 1989, 39-45). Interpretation is difficult and the internal rooms may have been used as quarters for family, servants, or travelers. Some structures were apparently bams for livestock. They were not common until the later second-century and may have housed the extended family (Hingley 1989, 39, 41). Burnham (1988, 44) classifies these as more
vernacular than Roman in character. However, it was not uncommon for some rooms towards the entrances to have several decorative features that were more Roman in style, indicating some aspiration to Roman culture (Hingley 1989, 45).
The more Romanized buildings, similar to the ones in larger towns or on villas, were relatively scarce in the small towns, and those that were present were generally found in the later periods (Burnham 1988, 45; Hingley 1898, 47-54; Burnham and Wacher 1990, 20). Many had hypocausts and other “Romanized” features or decorations such as painted plaster,
tessellated pavements, and mosaics, but most small towns only had one or two such buildings (Burnham 1988, 45; Burnham and Wacher 1990, 20). Public or official buildings were equally as rare (Burnham 1988, 55; Burnham and Wacher 1990, 20). Buildings associated with the cursus publicus {mansiones/mutationes) are often difficult to identify since there is little
epigraphic evidence to support their identification (Black 1995, 1). Basilicas were all but non
existent, with only one possible example being found at Carlisle (see Chap. 4). Only two
theaters have been identified, at Wycomb and Catterick. Two possible amphitheaters have been found at Charterhouse and Frilford (Burnham 1988, 55; Burnham and Wacher 1990, 20).
Bathing complexes, ranging from small simple to larger complexes, were better represented than would be expected as were religious buildings, though few were o f a typically classical style (Burnham 1988, 55; Burnham and Wacher 1990, 10).
The construction techniques used in Britain during the Roman era have been sufficiently covered in-depth elsewhere (see Burnham 1988; de la Bedoyere 1991, 15-39; Perring 2002, 80-
110; Burnham and W acher 1990, 16-23), that only a b rie f discussion is necessary here. Each has left its ow n unique archaeological traces which directly affect the data as it has been collected by excavators. Tim ber structures in particular are problem atic since w ood is m uch less durable and the traces that rem ain for the m odem archaeologist are m ore slight than stone, particularly before the m id-tw entieth century when archaeological techniques becam e m ore advanced.
* b
Fig. 2.2: Earthfast Construction (from Rosenheim 2000, 82) Image a is an example o f an earthfast upright in an individual hole; Image b is example o f an earthfast upright in a trench with both wattle and daub and timber planking.
Fig. 2.3: Horizontal sill beam construction with wattle an d d a u b (from P erring 2002, 89)
The most simple and cost effective tim ber structures w ere those w here vertical stakes or posts were placed directly into the ground, either in individual holes or a trench, w ith stones or earth packed around them to provide support. The supports could be filled either by m ud bricks or w oven w attle sails covered w ith clay daub (Burnham 1988, 39; Perring 2002, 83-4; see Fig.
2.2). The technique rem ained popular at low er status sites but was atypical in larger tow ns and cities (Perring 2002, 86). Tim ber structures could also be erected by placing a horizontal tim ber in a trench or on the ground surface w ith vertical posts held in place by notches and fram ing techniques. The super-structure could be mud brick fill, w attle and daub, or planking (Burnham
1988, 39; Perring 2002 87; Burnham and W acher 1990 17). This technique becam e more popular after the conquest when the R om ans introduced better fram ing techniques and showed an increase in expenditure (Perring 2002, 91; see Fig. 2.3). A nother im provem ent included the construction o f tim ber buildings w ith a stone foundation that w ould m inim ize the ground m oisture absorbed by the tim ber placed in or on the ground, dram atically extending the life o f
the structure (G oodbum 1992, 192; Perring 2002, 91). This was also accom plished by raising the tim ber superstructure on leveled stones, such as at W anborough (Burnham 1988, 39;
Burnham and W acher 190, 17).
• S . ^ '
Fig. 2.4: M asonry Construction (from Adam s 1994, 86)
Stone buildings indicate a higher level o f investm ent and w ere particularly popular in public buildings and aristocratic houses (Perring 2002, 91). The most com m on technique consisted o f a rubbled concrete core w ith stone facing (Perring 2002, 108-9; De la Bedoyere
1991, 25-6; see Fig. 2.4). The use o f larger blocks to create a solid stone wall, often referred to as ashlar m asonry, was extraordinarily rare (de la Bedoyere 1991, 25). Stone was generally acquired locally to m inim ize the cost o f construction (de la Bedoyere 1991, 24). H ow ever, m any small towns imported stone, often great distances, to construct buildings. As we shall see, the level o f investm ent w ould affect the m eaning o f each building.
The chosen construction techniques create certain problems for the archaeologist. In trying to quantify the transition from wood to stone over time it is important to gain as complete a picture as possible o f construction techniques across space and time. However, buildings that used stone, or at least had stone foundations, were far more likely to survive and be recorded, since they incorporated more durable materials (hence a reason to build in that manner). Many structures were robbed for their stone in the medieval period, however, these robbings were incomplete in many cases and even the more complete robbings left traces o f the masonry in the form o f robber trenches, masonry fragments, and foundations.
Wooden structures, on the other hand, were less likely to survive because wood decomposes easily and often leaves only ephemeral traces. In some rare cases water-logged timber has survived, such as at Carlisle, however, these are the exception, not the rule. In most cases only colorations in the soil remain to give any indication o f a wooden structure. Early archaeologists, because o f the state o f excavation methods and interest, often missed and unknowingly destroyed these traces and any meaningful data that could be derived from them.
In addition to archaeologists’ unintentional destructions o f these traces, the Romano-Britons could also easily destroy traces o f earlier timber structures when they built latter buildings, often stone as we shall see. Hence, we have a very incomplete picture o f the wooden buildings at many sites. On the other hand, there is no doubt that across the ages that timber has been the main vernacular building tradition in many areas o f Britain (de la Bedoyere 1991, 22; Ching
1995, 135; Potter and Johns, 2002, 100).
While the problems o f the buildings themselves are problematic enough, they are further complicated by the fact that not a single “small town” from Roman Britain has been completely excavated, much less done so to modem scientific standards. Some are more excavated than others, giving a larger number o f sample buildings than those that are less excavated. For example, the town o f Alcester has 71 building samples excavated, and Middlewich, Mildenhall,
and Ancaster have only one. This creates problems with statistical comparisons o f towns such as T-tests. It should, however, be noted that even sites with large numbers o f building samples are not necessarily without problems. For example, the town o f Corbridge has yielded 65 building samples. However, because o f the accessibility o f the site, it was subject to much excavation early in the twentieth-century when archaeological methods were more crude.
While much o f this evidence is still valid, it prevents a more complete picture from being obtained than a smaller site like Asthall, where excavation samples are smaller (20 building samples) but done with more modem methods employed over a smaller area.
Other factors create problems in the quantity o f data from certain sites. Later medieval and modem developments on the site in addition to archaeological interest and methods have greatly affected the data available today. The crude methods o f excavation, as mentioned above, have destroyed a significant amount o f data. In addition, early archaeologists tended to focus on stone buildings. In other places, places like Corbridge, that had no later medieval or modem development directly over the Roman settlement, excavations could be more easily done than places like Bath where later medieval and modem towns disturbed and/or sealed the Roman layers and limit excavations. While we have over 20 statistical samples from Roman Bath, it is tantalizing to contemplate what is as yet unknown beneath the modem town.
Excavations in living towns also create extreme variability in the methods employed to gather the sample data. Excavations often are limited in time and money, literally rushing to save what data is possible before modem construction destroys the site. Even in more academically driven excavations, methods vary, giving a complexity o f data that makes it difficult to compare one site with another.
To summarize, the complexity o f the data makes it difficult to carry out a purely
statistical analysis. Reece (1995) has examined the problems o f taking the varied data acquired by different methods at different times and evaluated their validity in “classical statistics.”
While the problems are very real and give true statisticians more problems than archaeologists, adjustments can be made in methodology to accommodate for these problems, though these adjustments are “not the stuff o f simple classical statistics” (Reece 1995, 180).