4. Database Construction

Data used in this study were sourced from the 1992 publication La necropoli laziale di Osteria dell'Osa (Bietti Sestieri 1992a). Each burial in this catalogue is described with the following data, where applicable (see Bietti Sestieri 1992a, 551-2):

Burial number
Alphanumeric designation to location on a 3m x 3m grid reference on site plan (Bietti Sestieri 1992a, pl. 3)
Latial Phase (see Table 1)
Damage to burial
Excavation date
Stratigraphical relationship to nearby burials
Double burials recorded in same grave
Ritual (inhumation or cremation)
Cluster name
Sex (indicating whether skeletally determined by Becker and Salvadei (1992), or if determined by grave good association)
Age in years (as identified by Becker and Salvadei 1992)
Age class (see Bietti Sestieri 1992b, 103)
Dimensions of grave (length, width and depth or diameter)
Orientation in relation to magnetic north
Location (X Y coordinates, expressed in centimetres, on an unpublished 1:200 plan of the cemetery (described Bietti Sestieri (1992a, 551). These coordinates are not truly geo-referenced)
Elevation of highest and lowest points of grave
Description of grave covering and fill, including description of location of grave goods in relation to body and/or grave boundaries
Detailed list and description of corredo

Each catalogue entry is accompanied by a planimetric illustration of the grave indicating the boundaries, skeletal material and grave good positions. There is also a section drawing of each grave, and illustrations of all surviving grave goods. There is a series of plans of the excavated area at the scale 1:200.

Although this catalogue is comprehensive, wide-ranging and objectively presented, there are a number of issues that need to be considered when using any such legacy data.

First, the translation of material qualities into published data necessarily entails a certain level of exegesis - decisions about the importance, categorisation and presentation of data affect the completeness and integrity of the dataset. In the case of Osteria dell'Osa this issue is manifested in the arrangement of burial information according to identified kinship groupings. This is easily overcome through the construction of an independent database, but observers dealing with the data as they are presented need to be aware that burial groupings are interpretative, not objective.

Second, the categorisation of objects within a standardised scheme is an inescapable problem. Material culture is meaningful only when it is named; such names are never objective, but engender the observer's interpretation of the objects they describe. In cases where our data are drawn from published sources we engage with the material via its interpreted identity; for example, if an object is described as a spindle whorl we must accept this identification unless we are able to examine the physical object. Even with such an opportunity, however, we cannot be sure of correct identification: objects are classified according to current functional understandings. These can be based on misguided intuition, and may be incorrect. It is important to remain aware of this significant, and inescapable, problem (this issue is explored in depth by Hodder and Hutson 2003, esp. 237).

The first step in spatially representing data from the cemetery of Osteria dell'Osa was to build a database that could be easily interrogated and exported to a GIS application. Initially a process was explored that used optical character recognition software (such as ABBYY Finereader) to read printed data and allow conversion to .csv or .xls format. This was used very effectively by Allison et al. (2004), but was not viable for the current project owing to the descriptive nature of the catalogue. Rather, it was determined that the most effective, although time-consuming, way of transferring the printed data into a useable numerical form was to manually populate a Microsoft Access database designed to accommodate a huge range of data in a standardised format that can be interrogated.

Very few data were directly transferable into this database due to their descriptive character. To avoid highly complex multivariate analyses that were deemed unnecessary for the current purpose, it was necessary to reduce these detailed descriptions somewhat to one of three formats: basic presence/absence, measures of quantity, or style classification (where this was both well defined and pertinent to the study). This process clearly omits potentially important detail, but could not be avoided given the scope of the project. Where more detailed data were later recognised as necessary (such as the size of finger rings), these were added, either as an adjunct to the main data table, or as a purpose-built linked table.

As construction of this database preceded analysis, decisions about what to include, and at what level of detail, were made in advance of knowing the full significance of each datum. As a result, some data that eventually proved extraneous were included in analysis and a need for others that had originally been omitted was identified as analysis progressed.

PREVIOUS NEXT CONTENTS HOME