Figure 3 shows the improvement in weight as a function of iteration number for the best chromosome in the population, satisfactory or not, with one line per experimental run. This is not to show the individual runs, but just to give a sense of how the algorithm improves solution quality over time. The dynamic algorithm starts out with an effective threshold of over 100 kips, so it is initially considering chromosomes that the threshold algorithm described above, with a threshold of 50, would not have considered. On the other hand the dynamic algorithm quickly (almost always within 30-40 evaluations) converges on a threshold of less than 50, and as the overall quality of the population improves the dynamic algorithm continues to focus on subsets corresponding to smaller and smaller thresholds.