Very nice write-up, Charlie!
As Charlie's statistics write-up points out, velocity SD is useful for estimating probabilities that single shot will fall within a particular velocity range. From that, you can make predictions of what effect the velocity variance of a load will have in terms of POI on the target (high/low). However, from any given data set, POI predictions made using SD will never fall outside (high/low) those made using ES. The difference is that ES does not allow you to make any estimate of where the in-between velocity shots will fall, whereas SD does.
Realistically, very few shooters ever utilize the complete information contained in their velocity data, which was the point of Charlie's write-up. Most simply develop a "feel" for whether a given load's velocity numbers are acceptable using either ES or SD as a measure of expected vertical spread due to velocity variance. I have always used ES, but the truth is that either ES or SD will work perfectly well for this simple purpose. If you wish to be more predictive with your velocity data, the SD value contains more information than ES and can therefore be used to predict vertical POI probabilities for all the shots in the string, rather than just the highest and lowest.
For F-Class shooting with 20-shot strings of fire, the reality is that a 20 fps ES will generally not produce [theoretical] vertical dispersion large enough to put you out of the X-ring at 1000 yd. However, if you're measuring velocity ES/SD from 5- or 10-shot strings, chances are very good your actual numbers on a longer string under match conditions will be significantly higher simply due to the larger sample size and barrel heating. For that reason, it's common to see statements to the effect that keeping ES at or under 10 fps, or SD in the low single digits, is what you want to see for a good F-Class load. Although such parameters can never rule out the effect of a single velocity outlier, they generally are a valid guideline for load development. The advent of e-targets and their resultant velocity measurements at the target face, as well as the use of a LabRadar for measuring long strings of fire shot under match conditions, may make it worthwhile for a larger number of shooters to analyze their velocity data in more detail.
I tend to view the whole velocity issue from a slightly different perspective. In my hands, it is not too difficult to keep the ES values for a good .308 F-TR load in the low teens or even slightly under 10 fps. For .223 loads, the values typically fall in the upper teens to low 20s. That is about the best I can routinely do without going to extraordinary lengths at the reloading bench. As noted above, those values will [theoretically] give me 10-ring vertical at 1000 yd, if not X-ring vertical. Past that, I really don't want to think about velocity numbers while I'm shooting in a match, because I can't do anything about it at that time. As long as I keep the test values where they need to be during load development and validation, I have confidence the values will also be acceptable during a match. In contrast, a 1 mph wind is enough to put me out of the X-ring at 1000 yd if holding center. Likewise, a 2 mph wind will put me into the 9-ring. As long as I keep my load velocity numbers at the best level possible I can during development, the wind will be by far the major source of error in my scores during a match. For that reason, I pay due diligence to ES/SD values during load development, and routinely carry out velocity (and temperature) measurements during practice sessions to ensure no major changes have occurred. Otherwise, I'm far more concerned with the flags and mirage because I know that is where I have the greatest chance of dropping points during a match.
