Internet Archaeology 4. Takács and Bartosiewicz-Quantitative Analysis

3. Quantitative analysis

In addition to the descriptive study of the geographical distribution of the identified species, potential bias in the interpretation of fish finds was studied from a purely quantitative point of view. The relationship between taxonomic richness (that is the number of fish species identified: R_f) and assemblage size (the number of identifiable fish and other animal bones: N_f and N_t respectively) has been reviewed in terms of regression analyses as applied by Grayson (1984: 138).

Results of the regression analysis listed in Table 2 cast some light on the trends chosen for study. According to the first equation, a degressive curvilinear relationship exists between the number of fish bones and total number of animal remains recovered at the 69 sites. Aside from the random bias generated by consumption and disposal practices at the various settlements (as is shown by the moderate but significant correlation between the variables: P<0.001), this trend may also indicate that at major excavations less attention is paid to the collection of small finds by hand (Figure 8).

Figure 8:Relationship between the total number of bones (Total NISP) and the number of identifiable fish bones
Fig.8 Relationship between the total number of bones (Total NISP) and the number of identifiable fish bones (Fish NISP) in hand-collected assemblages from Hungary.

The second function in Table 2 shows an even more strongly degressive tendency: the proportion of additionally identified new species decreases with the increasing number of fish bones recovered. This positive relationship is also highly significant (P<0.001). Curiously enough, the 0.241 exponent is only slightly smaller than the value obtained for the total number of species per total number of identifiable bones at 18 Roman period sites in Hungary (b = 0.277: Bartosiewicz 1989). These exponents fit the range of values for mammalian assemblages as listed by Grayson (1984, 144). This relationship thus may represent a general tendency which is relatively less dependent on recovery techniques as long as a reasonable proportion of the bones can be identified (Figure 9).

Figure 9: Relationship between the number of identifiable fish bones (Fish NISP)and the number of fish taxa identified
Fig.9 Relationship between the number of identifiable fish bones (Fish NISP)and the number of fish taxa identified (R) in hand-collected assemblages from Hungary.

Finally, a linear relationship could be established between the proportion of fish bones in the total assemblage and the number of fish species identified. While studied data contained only 3.5% fish bone as an average, an approximately 0.12% increase in the contribution of fish bones is needed to produce the third fish species in the faunal list. This not only indicates that the number of fish bones recovered by hand is hopelessly low in relative terms; but it also signals problems with the fish bone identification process. The moderate linear correlation between the variables is indicative of the variability of the taxonomic level to which bones were identified (Figure 10).

Figure 10: Relationship between the proportion of fish bones to total number of bones (Fish NISP/Total NISP) and the number of fish taxa identified
Fig.10 Relationship between the proportion of fish bones to total number of bones (Fish NISP/Total NISP) and the number of fish taxa identified (R) in hand-collected assemblages from Hungary.

Dramatic improvement brought about by water-sieving is well illustrated by the example of Keszthely-Fenékpuszta. Copper Age fish remains (not included in Table 1) were recovered in thousands using a 0.8mm mesh size. Although the analysis of the material has not yet been completed, the list of species identified so far is almost as long as the comprehensive list of species listed in Tables 1/a (prehistoric sites), 1/b (Roman sites), and 1/c (medieval sites). In addition, largely due to the successful recovery of many small pharyngeal teeth, two species previously undiscovered at Hungarian archaeological sites were identified: knife (Pelecus cultratus L.) and rudd (Scardinius erythrophthalmus L.). Results from this site indirectly support the tendency suggested by quantitative evaluation and provide a broader basis for the reconstruction of fish exploitation (Table 3).