Statistical documentation

A logical assumption for the assessment of the production of different categories of vessels is that the wall thickness for smaller vessels and fine-ware vessels used e.g. as table ware would be thinner compared to larger and heavy-duty vessels i.e. storage vessels. Thus dimensional and functional variations in the material could be discerned by a statistical ordering of the data using sherd thickness as the base parameter. With the help of simple statistics - such as mean based on the weigth of the classes, sample standard deviation and variance- the pottery data is transformed into a mathematical profile of the material analysed. This method of registration is best suited to materials displaying a relatively homogenous fragmentation of the pottery. Bottom sherds, handles and large vessel-parts may cause bias in the statistical description. It may be necessary to exclude these from the subsequent analysis.

In stead of using weight and number of sherds as a unit, an adjusted weight approximating the area of the sherds and reducing the bias towards thicker and heavier sherds, can be calculated (Hulthén 1974). Provided the density of the sherds in the material is roughly the same, and using 10 mm as a standard thickness all values for the sherd thickness classes are recalculated thereby upgrading the amounts of material below 10 and decreasing the amounts of material above. The result gives a more true picture of the relative volume of thin-walled vessels compared to sturdier built pots.

Formula for arithmetical mean
T=sherd thickness (mm); Wa=sherd weight (g); Wb=total weight (g)=SWa; Xt = mean sherd thickness

The distributions of sherd thickness calculated on number and weight of sherds or adjusted weight; of tempering types (type, amount and max. gr. size) in relation to sherd thickness, in relation to surface treatment or to vessel-size and type are some of the relationships, that may be studied by the use of descriptive statistics. The results are first and foremost presented as histograms on real amounts or calculated percentage, which facilitates a visual interpretation and comparison. The primary aim is to test the homogeneity of the material and to sort out different types of ware and their relation to and possible causation by differences in vessel forms and functions. A homogenous sherd material containing both small, medium-sized and large vessels will display a near-normal, unimodal distribution of the material across the sherd thickness classes. Compared to a true normal distribution the apex is usually shifted towards the thinner sherds in relation to the median. This profile of the material is not directly translatable into sets of vessels partly because of the internal chronological spread, which is always present in archaeological materials; partly because of the fragmentation, which makes it difficult to estimate the size of the pots and leaves some ware groups without sherds characteristic of the original vessel shape.

Formula for adjusted weight
Wt=true weight (g); T=sherd thickness (mm); 10=standard thickness (mm); Wa=adj. weight

Fragmentation expressed as the mean size in cm2 of the sherds, is another parameter, which may be estimated on the basis of the information in the registration . It is primarily related to the length and duration of human activities at the site of deposition although the size and wall thickness of the vessels is also of some importance. Consequently the estimations of the degree of fragmentation is a good indicator for mapping areas of low and high activity on a site. Proper statistical analyses such as linear regression, first component analysis etc are used in special case eg to study the interconnection between temper quality and vessel size.

 


Page maintained by Anders.Lindahlgeol.lu.se
This page was last modified 2005-01-28 12:29
To the first page