Initial results from jackknifing the entire procedure gave partial estimates of optimal fishing mortality scattered around 0.5 per year. The sustainable catch was 108 000 tons and had a 62% coefficient of variance (CV) and ranged from 8% lower to 52% higher than the value based on all the data. Omitting the first year's data created an outlier with a 47% lower sustainable fishing mortality. When jackknifing only the input to the logistic population model, the optimal fishing mortality retained a CV of near 63%, but the CV of sustainable catch fell to about 19% (ranging from 8% lower to only 7% higher). Much of the variation in estimated sustainable catch therefore appeared due to uncertainty in the standardized CPUE series. However, omitting the first year's data continued to give an outlying point, so the effect of leaving out the first year's data was investigated further.

When the experiment was repeated with the first two data points permanently omitted, the estimate and uncertainty of MSY decreased to around 82 000 tons with a CV near 18%, while the variation in sustainable fishing mortality and in standing stock increased. Also, there was little difference whether the jackknife procedure included the General Linear Modelling fitting of the CPUE series or whether points were simply omitted from a single series based on all the years' data. Thus particular properties of the data points appeared to have significant effects.

The jackknife procedure appears useful for examining the reliability of this standard method of fishery assessment, and these results indicate that the uncertainty of such results may be unpredictable, large, variable between model parameters, and difficult to assign to any particular step in the procedure.