There are at least a couple major caveats in regard to the answers to your questions. First, as you change seating depth, you will change the effective internal case volume, thereby
potentially affecting pressure and velocity. The real question is how much can you alter seating depth before pressure and velocity are noticeably altered? If you change seating depth by a sufficient amount, the velocity
can change. In my hands, changing the seating depth of a
jumped bullet (i.e. a bullet
not seated into the rifling) by .010" to .015" in either direction may typically
not be enough to cause a noticeably change in velocity, or at least one that can accurately be measured by a chronograph. In other words, you may detect what appears to be very small velocity changes shot-to-shot, but the average velocity change between a 5-shot string at one seating depth and another will be less than the standard deviation, and therefore is not effectively a statistically significant change. However, changing seating depth by ~.050" in either direction can easily cause a noticeable and
statistically significant change in velocity. So the total range over which you test seating depth is one caveat.
The larger that range, the greater the chance that the velocity data will change. Further, if you change seating depth by an amount sufficient to change the average velocity by a certain margin, it is entirely possible that you could put the load out of the optimized charge weight window.
For example, Berger Bullets has put out a method for optimizing seating depth with VLD bullets, which are sometimes finicky with respect to seating depth, that covers an extremely wide range in very coarse increments:
https://bergerbullets.com/getting-the-best-precision-and-accuracy-from-vld-bullets-in-your-rifle/. If covering such a wide range during seating depth testing, it is practically guaranteed that velocity will change noticeably, possibly moving the load out of the optimal charge weight window. In such an event, it would be necessary to re-visit charge weight testing within the much narrower optimal [coarse] seating depth window, before proceeding to fine-test seating depth within that narrower window in small increments. To be clear, this sort of "which is first, the chicken or the egg?" testing results are not uncommon in the reloading process. Any time your start testing too far away from what ends up as the optimal region for a given variable, it is likely you will need to re-visit that variable again once you actually optimize a different variable. The good news is that you can readily test charge weights, move to seating depth optimization, move back to charge weight testing for fine-tuning, then optimize seating depth some more. The order in which you carry out the steps isn't as critical as the final result. This kind of incremental or stepwise testing is not at all uncommon when trying to optimize multiple variables.
Another caveat lies in the group size used to determine ES and SD. For example, if you're using 5-shot groups to determine ES/SD, the values obtained can easily change from day to day on different range trips, due to the small sample size (n). A load that gave an ES/SD of 15/7 fps one day night give an ES/SD of 25/12 fps (or greater) on a different day. In other words, what you thought were nice low ES/SD values actually weren't. Again, that can happen because the sample size was so small the numbers weren't statistically significant. The more times you obtain consistent ES/SD measurements for a given load on different days using small sample sizes as the velocity data, the more reliable those numbers probably are. In fact, if one were to shoot 10 x 5-shot groups with the exact same rifle/load on a single day, making absolutely certain the rifle/barrel completely cooled between individual shots, as well as between groups, I suspect many might be surprised at how large the range of ES/SD values they would obtain might be.
To be clear, I am not advocating shooting 10 x 5-shot groups just to estimate ES/SD values. Under such circumstances, typical load development would consume far too many rounds of usable barrel life. However, one needs to be aware of the the inherent issues associated with using small samples for statistical analyses. There are some things you can do to realistically achieve confidence in your ES/SDS values during the load development process without necessarily shooting the barrel in the process. As I mentioned, compare the ES/SD values obtained using a small sample size for the exact same load on different days, just to convince yourself the nice low values you obtained for a single 5-shot group a few days prior are still holding true. Second, some F-Class shooters follow a load development regimen that consists of essentially three parts: charge weight testing, seating depth testing, and finally, load
validation. Arguably, load validation may considered a different animal than load development; nonetheless, I'm lumping it in with the load development process for the purpose of this response. F-Class matches typically consist of several long strings of fire (25+ shots) throughout the course of a morning/ear/y afternoon. Because of the long strings of fire and the fact that the atmospheric conditions usually change during the course of even a one day match, low ES/SD values are generally considered highly desirable.
So let's say you've identified what looks like a good optimal charge weight using 5-shot groups. You then optimize seating depth using that charge weight, meaning however many more 3-shot or 5-shot groups you need to cover the chosen seating depth test range. As I noted previously, it is possible to change seating depth of a
jumped bullet by at least .010" in either direction without substantially altering the average velocity. For that reason, you would then expect the low ES/SD values previously observed for the 5-shot group of your chosen charge weight during the prior testing step to hold up across several adjacent seating depth increments within your seating depth test range, if not the entire range. Remember, if you test across much larger seating depth increments, the velocity data likely will change. Nonetheless, the velocity data from adjacent segments of the seating depth test can be used to further support [or refute] the notion that your chosen charge weight appears to provide stable and reproducible [low] ES/SD values.
OK. So you've now identified a suitable charge weight and optimized the seating depth to tighten up the groups. So far, the velocity data are looking good, even though there is still the caveat of small sample size. So now the idea is to
validate the load. For an F-Class shooter, that might involve loading up 25 rounds or so of what they believe should be the optimized load (i.e. specific charge and seating depth). Next, they would shoot several sighting shots, followed by 20 shots for record, just as they would do in an actual match, trying to maintain shooting cadence, focus, attention to conditions, i.e. do everything just like they would in an actual match. The only difference is that they're also recording velocity data. In other words, they are attempting to
validate that everything their preliminary testing has suggested to be true about the load actually holds up under match conditions. If the load doesn't hold up in terms of both velocity and precision, further testing is warranted. What is "good enough" or "sufficient" can only be determined by the shooter. Sometimes, these multi-step charge weight/seating depth/load validation approaches can be painful. In that event, I always try to remember that at least a few of my competitors will almost certainly
not be skipping any steps or cutting corners, which usually provides sufficient motivation to continue.
