16 
Fishery Bulletin 95(1 ), 1997 
Table 3 presents the resulting values for the rela- 
tionship between the greatest centrum length, GL, 
and greatest proximal centrum breadth, GB(p), to 
the skeletal length, SL, for the recent specimens as 
calculated by regression analysis. In this table, some 
values are missing or are based on only 3 specimens 
owing to the original preparation of the comparative 
skeletons. Standard deviations and confidence lim- 
its are not given but are available on request. 
Size determination of archeological 
specimens 
The standard formula used to estimate the size of 
fish represented by the archeological specimens is 
given by Casteel (1976, p. 96) as: log (fork length) = 
a + b x log (GL or GB). The constant (a) and the 
slope — or x coefficient — (b) are taken from Table 3 
(i.e. the values derived from the comparative speci- 
mens), and the logarithm of the greatest length, GL, 
or proximal breadth, GB(p), from each archeological 
specimen (Table 4). 
Definitions of vertebral measurements taken from both 
comparative and archeological specimens of bluefin tuna. 
Greatest length (GL): maximum length of the centrum, 
taken at the lateral midpoint with digital calipers and 
measured to the nearest mm. Greatest breadth (GB) = 
maximum breadth of the centrum, taken at the lateral mid- 
point of the proximal face, GB(p), and distal face, GB(d), 
taken with digital calipers and measured to the nearest 
mm. Radius (R) = the maximum distance from the center 
of the cone to the edge, of the proximal face, R(p), and dis- 
tal face, R(d), taken at the lateral midpoint. This measure- 
ment was taken with a plastic ruler cut diagonally to fit 
into the cone of the centrum; in this way the amount of growth 
from the center of the cone to the sharp raised ridge at the lip 
of the centrum was measured to the nearest 0.5 mm. 
Casteel (1976) noted that although the regression 
method is the most accurate way to estimate fish 
length from bone size, these length estimates always 
vary somewhat between vertebrae from the same 
individual, even when the predictive value (r) of the 
equation is high. When both length and breadth 
measurements were available for an archeological 
specimen, the measurement that produced the length 
estimate with the highest correlation coefficient (r) 
value for that individual was used to represent that 
fish. Alternatively, an average of all available mea- 
surements could have been made, although this 
method allowed both comparative and archeological 
specimens to be treated similarly. 
The method used in the present study required that 
vertebral specimens be identified to exact column 
position. This can be problematic for archeological 
specimens because several of the centra in the ver- 
tebral column are almost identical and because ar- 
cheological specimens may often lack diagnostic neu- 
ral or haemal arches and spines. However, an archeo- 
logical specimen can almost always be defined to a 
small range within the column (e.g. vertebrae num- 
bers 14-16). Vertebrae not identified to exact posi- 
tion were found to be so similar in size and propor- 
tion to adjacent vertebrae that they could be treated 
as interchangeable for the purpose of the estimations 
attempted here. Where the exact position of an ar- 
cheological specimen was uncertain (which occurred 
for less than one third of the specimens examined), 
the number of the vertebra used to calculate the size 
estimates is given in parentheses, e.g. (15). 
Table 4 presents all archeological vertebrae mea- 
sured (by vertebra number) and the length estimates 
derived from them. Where eroded edges prevented 
accurate measurement, an estimate was taken if it 
was likely to be accurate to within 1 mm. A total of 
78 vertebrae were measured, representing at least 
45 individuals. Several vertebrae were found at- 
tached (occasionally in articulated position) or could 
potentially have belonged to the same individual by 
virtue of similar size and proximity within the ar- 
cheological deposit (this is a standard assumption 
for determining the minimum number of individu- 
als represented by skeletal remains recovered from 
archeological contexts). Radius measurements of 
these specimens were also taken (because this di- 
mension is preferred by some researchers for ageing 
purposes) but are not reported or used in the calcu- 
lations. All measurements are available on request 
from the author. 
The fork-length estimates for the archeological 
sample listed in Table 4, as for the comparative skel- 
etons, are derived by adding 30 cm to the estimated 
skeletal length to yield a fork length (to account for 
