Pottery and tile - detailed analysis

5.2.1 The pottery and tile were examined, classified and quantified

Pottery and tile fabrics were analysed, given fabric numbers and assigned to chronological periods associations by Steven Willis. A summary of these fabrics is provided in Tables 5.1 (pottery) and 5.2 (tile).

Table 5.1 Summary of pottery fabrics. (For quantities found, see Table 5.9)
Links go to fabric description

Fabric Hand-made Wheel-made Wheel-turned Chronological phase
101YesNoNoIron Age
103YesYesNoIron Age
104YesNoNoIron Age
105YesNoNoIron Age
107YesNoNoPossibly prehistoric
110NoNoNoPossibly Iron Age
201NoYesNoPossibly Roman and/or later
204NoYesNoPossibly Roman and/or later
205NoYesNoPossibly Roman and/or later
208NoNoNoNot in use
215NoYesNoPossibly modern
304NoYesNoPossibly modern
307NoYesNoPossibly Roman and/or later
308NoYesNoPossibly Roman and/or later
309NoYeNoPossibly Roman
313NoYesNoPossibly modern
314NoYesNoPossibly Roman
319NoYesNoPossibly Roman
322NoYesNoPossibly Roman
325NoYesNoPossibly modern
326NoYesNoPossibly modern
329NoYesNoPossibly modern
402NoYesNoPossibly Roman and/or later
403NoYesNoPossibly Roman
601YesNoNoPossibly prehistoric
603YesNoNoPossibly Iron Age
604NoNoNoIron Age

Table 5.2 Summary of Roman tile fabrics. (For quantities found, see Table 5.11)
Links go to fabric description

Fabric Evidence of use
for tegulae
Evidence of use
for imbrices
Other sites where
fabric has been noted
1005YesYesVizela (tegula); Braga (tegulae)
1006YesYesBraga (tegulae)
1008YesYesSanfins; Terroso; Vizela (imbrex); Braga (tegulae)
1012YesYesTerroso; Vizela (tegula)

5.2.2 The quantities of each individual pottery and tile fabric were converted to values per ha.

The quantities of each tile and pottery fabric were measured by sherd numbers and weight, before both sets of values were standardised as quantities-per-hectare. Since individual fields were used as the collection units, and thus as the sample areas on the distribution maps, the number of sherds of each fabric recorded from each individual field had to be standardised in relation to its area. The following formula was used:

S = (n x 10000) ÷ A   where S = sherds per hectare
n = number of sherds
A = field area in m sq.

The field areas were calculated within the GIS in ArcView, and the formula was executed on the sherd numbers and weights for each field using Excel.

Query pottery quantities by field | Query tile quantities by field

5.2.3 The dated pottery was grouped by fabric into broad chronological phases for analysis

On the basis of the pottery analysis, each fabric for which there was evidence was either assigned to a broad chronological phase, or described as 'possibly' linked to a particular period (Tables 5.3 and 5.4). The ceramic phases used are: Iron Age; Roman; Medieval. Fabrics of less certain date were assigned as follows: Possibly prehistoric; Possibly Roman; Possibly Roman and/or later.

Table 5.3 Summary of fabrics of known date

Iron Age fabrics Roman fabrics Medieval fabrics
101102 311
103 108 312
104 303 316
105 305 501
604 305A 

Table 5.4 Summary of fabrics of uncertain date

Possibly prehistoric fabrics Possibly Roman fabrics Possibly Roman and/or later fabrics
107103 201
110 309 204
601 314 205
602 319 307
603 322 308
  403 402

5.2.4 The median and percentile values were established for each chronological group, and also for selected individual fabrics

The ranges of data values for the pottery densities in the walked fields do not conform to a statistically Normal Distribution. Consequently it was decided to map the density plots using scales derived from the median value and percentile bands, as opposed to the arithmetic mean and deviations from it. The median is the middle value of a set of data that has been ranked in decreasing order of size. In this instance, the median is the middle value of the ranked data from 855 fields, i.e. the 428th value. For the production of density scales, the ranked values were also sub-divided into 10% blocks, or percentiles. The threshold of the tenth percentile thus represents the ranked value 85th from the highest value in the data, the threshold for the ninth percentile is the 171st from top, and so on.

The densities of pottery and tile contained within the top percentile bracket for each group are thus the highest found, and are thus the most likely to be of archaeological significance. This method of scaling has the advantage of allowing comparison of fabric and period distributions of pottery based on the overall distribution of a particular fabric or group of fabrics, regardless of absolute quantities of pottery in the study area. Although this method allows comparison of high densities of pottery for different phases, any assessment of the significance of such concentrations must be based on an examination of the material recovered and its landscape context.

Median and percentile values were established for phases and fabrics using Microsoft Access and Excel. Both sherd numbers and weights of pottery per-hectare were used to provide a comparison between respective densities for each phase or fabric.

Query pottery fabrics per hectare | Query tile fabrics per hectare

Visual comparison of ranked quantities of sherds of specific phases, together with initial analysis of pottery densities spatially, led to the decision to provide an additional value at the mid-point of the top percentile of each group, thus separating the top 5% of the values from the remainder. These values were calculated because of the low absolute densities of Iron Age and medieval pottery which meant that all fields in which pottery of these periods had been found were included in the top percentile range (Table 5.5). This sub-division was thus designed to help us to assess the significance of fields falling within the top percentile.

Table 5.5 Percentile values of quantities of pottery and tile by period

Fabric/Period Iron Age pottery fabrics Roman pottery fabrics Medieval pottery fabrics Roman tile fabrics
Value type Sherd number Weight (g) Sherd number Weight (g) Sherd number Weight (g) Sherd number Weight (g)
Median value 0 0 0 0 0 0 0 0
Mean value 0.812 2.785 2.952 133.328 0.359 1.511 97.790 8199.240
Standard deviation 10.673 20.660 30.529 3183.086 1.355 7.730 2715.768 227706.688
Lower threshold of 7th percentile 0 0 0 0 0 0 0 0
Lower threshold of 8th percentile 0 0 1.039 2.715 0 0 0 0
Lower threshold of 9th percentile 0 0 2.798 10.268 0 0 1.086 29.760
Lower threshold of 10th percentile 0 0 5.796 33.058 0.846 1.742 6.046 350.246
Mid-value of top percentile 1.221 4.560 10.193 80.956 2.441 8.760 24.073 1469.559
Maximum value 294.118 322.038 882.353 92941.176 14.908 138.446 79411.765 6657941.176

5.2.5 Maps showing the density of the various pottery and tile distributions were created using the percentile values as scales

The median and percentile values shown in Table 5.5 were used as the scales for maps showing the density of pottery of the three ceramic phases and also Roman tile.

5.2.6 High density concentrations of pottery and tile relative to the background distributions were identified on these maps, allowing the definition of probable sites

The top percentile values thus defined the peaks of density in relation to the background distributions. Fields with values in this range were located on the maps using ArcView 3.1 software at Southampton University. Any clustering of fields with high ceramic densities was noted, together with any pattern of decreasing densities in adjacent fields. Such concentrations of material were encircled on the maps in ArcView and considered as potential sites).

| View newly-identified sites density plots |

5.2.7 The distributions of fabrics of uncertain phase were compared with those of the dated material in order to establish whether there was any correlation between them

The distributions of ceramics of known date were then compared with those of pottery which had been defined as 'Possibly prehistoric', 'Possibly Roman' etc. to assess whether they showed any spatial association. In particular, this involved comparisons between pottery of definite Roman date and that categorised as 'Possibly Roman', and 'Possibly Roman and/or later'. Fabric 308, which accounted for a substantial proportion of 'Possibly Roman and/or later' pottery was also compared separately with the concentrations of dated pottery (Tables 5.6 and 5.7).

Table 5.6 Percentile values of quantities of pottery of uncertain date

Fabric/Period 'Possibly prehistoric' fabrics 'Possibly Roman' fabrics 'Possibly Roman and/or later' fabrics
Value type Sherd number Weight (g) Sherd number Weight (g) Sherd number Weight (g)
Median value 0 0 0 0 3.233 10.220
Mean value 0.394 1.951 1.338 21.920 12.734 46.569
standard deviation 3.217 20.755 20.338 533.242 27.574 102.941
Lower threshold of 7th percentile 0 0 0 0 5.594 19.166
Lower threshold of 8th percentile 0 0 0 0 9.448 37.732
Lower threshold of 9th percentile 0 0 0 0 16.936 62.459
Lower threshold of 10th percentile 0 0 1.683 5.716 34.194 121.147
Mid-value of top percentile 1.611 4.729 3.625 18.096 59.709 204.769
Maximum value 62.677 500.816 588.235 15588.235 344.247 1324.721

Table 5.7 Percentile values of quantities of fabric 308

Fabric/Period Fabric 308
Value type Sherd number Weight (g)
Median value 2.159 6.279
Mean value 10.024 37.610
Standard deviation 21.692 89.561
Threshold value of 7th percentile 4.621 14.454
Threshold value of 8th percentile 7.868 30.622
Threshold value of 9th percentile 14.310 51.398
Threshold value of 10th percentile 27.526 102.437
Mid-value of top percentile 45.968 166.898
Maximum value 305.381 1324.721


© Internet Archaeology URL:
Last updated: Mon Dec 18 2000