The results for our “medium”-sized problem, with 44,928 members, were not as good but were still encouraging. Figure 4 shows these results for the same algorithms, and again for both the satisfactory and valid criteria. The scales on both axes have been set to match the previous graph for ease of comparison. A threshold of 150 was used for the threshold algorithm. As before, the dynamic algorithm with a selection factor of 30% shows the best results, and achieves most of the benefits it is able to achieve within relatively few evaluations. Out of sixteen runs, nine found satisfactory solutions within 250 evaluations. Weights ranged from 46.19 kips to 74.86 kips, with a mean of 62.31 and a median of 59.01. Prior to these runs our best solution was around 78 kips, so 46.19 represents a new best, and is consistent with a structure that is roughly four times larger than the smaller one for which the best known solution has a weight of around 10 kips.