How does a department make the transition from low to high rank?
Perhaps the appearance of a system undergoing a phase transition at a boundary is more than a coincidence. Most of the physics departments in the US began small with a preponderance of superatomic physics. Since the direction of a department is largely determined by a vote of the faculty and it might be in their perceived self-interest to strengthen existing research groups or develop similar ones during periods of growth, the preponderance of superatomic physics tends to be a highly stable phase.
Nevertheless many departments have made the transition. It would be interesting to interview faculty at these schools to determine how it was done. I suspect that in many instances there were one or two far-sighted faculty who realized that their department could attract more and better students and more research funding in all areas of physics if they moved to balance their research profile.
If we divide the data in the figure into angular regions, the ratio of high ranked programs to all programs in each region gives a measure of the probability of a program with such a research profile to be high ranked rather than low ranked. Clearly this probability is an increasing function of the angle. In fact there is a narrow range in the angle in which the transition from a low probability of being high ranked to a high probability takes place. The transition is even more rapid in the case of small departments (<32 faculty). The transition is shown in the attached figure. For a more quantitative study of the NRC quality ratings click here