Clusters. Cluster 1¹ of the protosegment *b and Cluster 2 of *B (representing the bilabial implosive in the bottom four lines - Wordcorr handles full IPA but our Web software doesn't) differ only in the third variety listed in the current view, so the two clusters contrast with each other in initial position.

Refine. Refine lets you look over all the results and notice things like this sort of contrast or complementation among clusters of correspondence sets. It often involves reordering clusters or assigning correspondence sets to a different protosegments than you you did when they first showed up, or changing the environment of some sets. The Refine panel contains buttons that do all these things and more:

Some correspondence sets, like *B Cluster 1, the fourth line from the bottom, clearly don't fit a regular pattern. You move them into Residue. They are still accessible and valuable; but they are more likely evidence for linguistic borrowing than they are for regular sound changes.

Each cluster is evidence for part of your analysis. By the time you're done, the clusters will be arranged so as to show the ways each protosegment has developed in each environment. Clusters that break the pattern will be preserved as residue.

Displays. The Summarize Evidence button pulls together all this in a display equivalent to the appendix of supporting data that appears in books and articles on historical linguistics. The Output Results button (coming in Release 2.1) enables you to print out all the information in the Results structure.

¹A cluster is a group of correspondence sets that share the same environment and differ from each other only in the number or placement of Ignore symbols ("."), which represent missing data and similar things. The clusters for each protosegment are numbered sequentially, and may be rearranged using Clus Order.