Turn experimental data into reviewed statistical analysis.
Recent revisions are kept. Restoring one creates a new revision.