Hi Andrew.
Good questions. Because the ochric is a "catch all" epipedon (for all surface layers that don't meet the criteria for any other epipedon), the rules have always been somewhat confusing. I'll try to answer your questions as best I can given the vagaries of the ochric definition. And as you point out, in the end there is little practical significance to most of this.
1) The phrase "appreciably darkened" has no quantitative definition. When the surface is "appreciably darkened" the person describing it can see the darkening (even if faint) and will call the horizon an A horizon. If not, she may call it an E or B or whatever the case is, right up to the surface. However, by convention, we say any mineral horizon that is plowed, (darkened or not), is an A horizon. So an E horizon at the surface, when plowed, becomes an A horizon. When ST was developed, 18 cm was a generally accepted estimate of how deep we plowed. So in the relatively rare instances where a soil has no A horizon, we arbitrarily say we'll identify an ochric epipedon in the upper 18 cm (as long as we have a subsoil of alteration as specified in the ochric criteria). I have personally only rarely seen soils like this.
2) In the scenario you provide, I believe the ochric includes the O plus underlying A horizons, however thick they are.
3) When soil taxonomy talks about mixing it limits it to 18 cm (because that is considered a normal plow depth). You don't mix deeper, but you do mix shallower if for example you hit a paralithic contact or similar layer before 18 cm.
4) Weighted average of colors. My short answer is that you should round to whole values because that is how we record them in the first place. Weighted averages are more useful for physical properties, like clay or organic matter content. because you will get a mathematical result pretty close to what you would get if you thoroughly mixed a sample of two layers in the correct proportion and then analyzed it in the lab. Since color may vary between ped surfaces and interiors, I'm not sure a mathematical weighted average is a good approximation of physical mixing. Sometimes yes, sometimes no. But I can understand why you are doing it as a way to look at your data.
