Figure 6

Regression of C _q against logarithm of dilution. Amplification efficiency is estimated from the slope of the line relating C _q to the dilution, the latter on a logarithmic scale. The confidence limits for the slope, when calculated from unweighted least-squares (as commonly done in simple linear regression) is strictly appropriate only for normally-distributed homoscedastic data. We caution that our data are not homoscedastic, and are unlikely to be normally distributed.