there are many reasons why this can occur...
- statistically insignificant sample size
- variable lighting affecting chrony
- bore condition, clean vs fouled
- barrel temp, ambient temp, ammo temp, powder temp sensitivity
- shooting position and inconsistent hold from shoulder fired weapon
- brass volume consistency
- seating depth
If you can't come up with anything logical, then see if you get repeatable results by alternating shots from each load. If after 30 samples each, the mean of one load is 2x SD apart, then you can have 95% confidence that the MV going forward will be about that much different.
Either way... how did the groups compare? How did POI compare?