Yes - statistical variation. You simply will not get the same results every time, even with a perfectly stable temp.
Apparently you misunderstood my method as well. I take Lum first, then the filter. This is done RIGHT AWAY after the Lum. No way would you do lum and then each filter, way too much time for things (mostly temp) to change..
2) Filter #1
4) Filter #1
6) Filter #1
8) Filter #1
10) Filter #1
The another 10 using Filter #2, etc
IMHO doing offset calculation just once is asking for trouble and my numbers clearly show this to be the case. Here are offsets for the blue filter taken as shown above, one right after the other with a stable temp:
In fact, that run was not too bad and I redid a couple that were larger outliers. If you just do one run and you hit one of those on a one time deal, you are gonna use a bad offset. Multiple runs is simply the only way to deal with this.
This is not a cheap or badly set up system, BTW. It is a Starlight on a TOA-130 (also FSQ) with the HSM motor.