The method involves two main steps. First, relationships between the size of certain fish bones and live fish length (and weight) are established for a particular species. This is achieved by the metrical analysis of a modern sample, usually consisting of 150-200 specimens of the species under study, and the generation of regression co-efficients which describe these relationships. Secondly, the archaeological material is measured, original fish sizes are estimated using the regression co-efficients, and dispersion statistics and size-frequency histograms are produced.
The bones used for measurement are the five paired cranial bones (premaxilla, maxilla, dentary, articular and quadrate), which have been used for many years to quantify prehistoric fish catches from archaeological sites in New Zealand, and the otoliths. As these bones do not always survive intact in archaeological sites, it is desirable to include measurements that are applicable to incomplete bones. The measurements are clearly defined, and closely parallel those used by archaeozoologists on other species.
Press here for an illustration of the measurements made on blue cod bones.
The measurements taken on the modern sample are entered into a computer database, and reliable regression relationships are established between bone dimension and live fork length. Regression analysis is carried out on the measurements of each bone dimension individually, and various types of curve fitting procedures are tested. A number of regression models are examined to work out the optimum estimator for each bone measurement.
Press here for a picture of blue cod dentary size plotted against live fork length.
Measurements are then made of the archaeological bones. Whenever possible, the largest dimension is taken, as this yields the most reliable estimate of the original fish size. A somewhat problematic issue arises with the selection of bones for measurement. Should the measurements be made on all available bones, or on a sample that is equal to the MNI for the site? This issue has been the subject of formal theoretical analysis by Leach and Boocock 1995, using a computer simulation model. This involved taking a large sample of bones from a fish catch where the size-frequency diagram and associated dispersion statistics were known, and carrying out recursive simulated breakage of bones so they could not be measured. It was concluded that measuring all bones of a species in order to arrive at a size-frequency histogram of the original catch does not introduce bias into the results, so long as the survival rate by bone size is random and the sample is of a reasonable size.
Measurements are then converted into estimates of fork length using a computer program, and size-frequency diagrams and dispersal statistics produced.
Press here for the size-frequency diagram for blue cod in the catch at Mana Island.
