Good practices for rational subgrouping:
1. Arrange your subgroups so as to capture as little variation as possible within subgroups. [A subgroup is a group with all one factor, in your Case "headstamp". What you talking about (i.e. groups of five) that is considered a “subgroup” size.] To do this, you may have to e.g. weigh each case to get as little variation for each individual subgroup, or make sure they are all the same trim length WITHIN EACH INDIVIDUAL subgroup.
2. Run your different subgroups (during your trials) to capture the MOST variation between subgroups. Determine what factors are going to determine the most variation" (e.g. cold Barrel versus warm/hot Barrel, cleanbore versus fouled). You can block these factors bye doing your trials before they have a chance to change. For example run all your treatments with a cold Barrel before it has a chance to heat up. In past practice, a cold Barrel versus a hot Barrel will have a big factor on where the point of impact is.
Remember during experiments, there is always bias; either conscious or unconscious. So if you know what each group comprises (that is, what headstamp is from that subgroup), you may subconsciously force one to be better or worse. Good luck.