Each structural pattern in the investigation covers a span of entries. But coverage alone doesn't show when the investigation was concentrating on that pattern — how many entries in any given stretch belong to it. This page computes a rolling-window density for each pattern: at each entry position, what fraction of the surrounding entries were tagged with that pattern?
The result is a kind of spectrogram of intellectual attention. Periods where a pattern runs dense appear as peaks. Gaps appear as flat lines. The chart shows where the investigation was converging versus dispersed, and which patterns rose and fell together.
Use the window control to change the rolling window size. Larger windows smooth local variation; smaller windows show sharper phase transitions. Click a legend item to isolate that pattern.