Fig. 9. Correlation between lipid content at the start of the trip, predicted from the fitted drift rates at departure (using the fitted spline function), for each individual seal and the lipid content measured just prior to departure using the labelled water method. The thick black line represents predicted lipid contents being identical to the measured lipid contents. The red circles (each circle representing an individual seal) were obtained using the drag coefficient (C_D) that minimised the sum of squared residuals between predicted and measured values (C_D=0.69), while the blue crosses (again, each cross representing one individual seal) represent the analysis using a C_D value that produced a regression slope of unity (C_D=0.49; see text and Fig. 8). The regression lines for these two models are represented by the red and blue broken lines, respectively (see text for regression results). Individuals for which the first daily fitted drift rate (and predicted lipid content) occurs more than 10 days after departure (20918_99, 28497_99 and 28504_99; see Table 3), and/or for which the initial fitted drift rate (and predicted lipid content) was calculated from one isolated drift dive early in the record followed by a long gap in the data (2846_99 and 26627_99) were excluded from the graph and the analysis. Also, one individual (FirstOne_00) for whom the drift dive record shows an unusual pattern, and is thus probably subject to significant dive misclassifications, has also been excluded.