So far, we have looked at two topic areas in physics, heat transfer and propelled motion. With heat transfer, we found that understanding of causal mechanisms showed a big improvement around 11 or 12, progress which was associated with a corresponding improvement in variable selection. As we probed further, the association between mechanisms and variables began to look like anything but an accident. On the contrary, there were strong indications of contingency, such that improvements in variables may have been dependent on improvements in mechanisms. As noted repeatedly, such contingency amounts to theoretical structure. With propelled motion, we found that understanding of causal mechanisms improved around 14 or 15, and that once again this was paralleled by progress over variables. However, this time, there was no evidence for interdependence. Rather, the indications were of two separate systems, one comprising the variables relevant to overall speed and the other the mechanisms by which speed change is ‘perceived’.