George,
I am working backward from what I see in table 310.16, not from any knowledge of how 310.16 was generated.
Take a look at the formula in 310.15(C) Engineering Supervision.
This formula is pretty darn simplistic. You presume an allowable conductor temperature rise (TC - (TA + {delta}TD) ) and divide it by the thermal resistance to the surrounding ambient RCA. This tells you how many 'thermal watts' you can dissipate, and is essentially 'ohms law' for heat. You have a certain amount of heat that flows, and this requires a certain temperature difference across a thermal resistance.
The amount of heat generated in the conductor scales as the electrical resistance of the conductor and the _square_ of the current flowing through it. This means that the current which can be carried at a given limiting temperature will vary as the square root of the amount of heat which gets carried away from the conductor.
If you presume that {delta}TD, RDC, YC, and RCA are _constant_, and presume that TC is the value (60C, 75C, or 90C) at the top of the columns in table 310.16, and then let TA vary, then you will find that the results given by the equation in 310.15(C) will vary in the same way as the correction factors at the bottom of table 310.16. This makes it clear to me that these correction factors were developed using this formula or a very similar one, and that they are based upon the conductor temperatures at the top of the column.
Reading across the columns, the allowed temperature rise doubles (from 30C to 60C). This means that heat carried away from the wire would double, and we thus expect the ampacity to increase by the square root of 2, or 41% more current. Actually looking at the numbers, however, we find that the ampacity increase from the 60C column to the 90C column is about 35%. However we have to note that the resistance of copper wire increases as it gets warmer, so that the resistance of copper at 90C is about 10% greater than that at 60C. 1.35^2 * 1.1 = 2, so again the formula in 310.15(C) fits the values in table 310.16.
This leaves me quite certain that the values in 310.16 are in fact consistent with the conductors reaching their temperature ratings when loaded to their full ampacity. The big assumption made by the table is the RCA value, which we are never asked to calculate or determine. Clearly the thermal resistance to ambient will be vastly different for conductors installed as knob and tube in un-insulated space, versus NM in an insulated wall. My guess is that in any situation where you can actually go and feel the conductors during use, the RCA value is far less than assumed by table 310.16.
Now... we've done all that. The wire ought to never exceed 60 degrees, as we have calculated. We next attach it to a breaker rated to withstand 75 degrees ... how can there be a problem?
If I have 60C rated 12ga wire, connected to a lighting fixture that draws 1A in a 30C ambient, then the conductor should never self heat above its temperature rating. Yet most new lighting fixtures require 90C wire. Why? Because the lighting fixture itself produces heat.
I would be _very_ surprised if a breaker were to produce enough heat to damage a wire. But if the breaker is marked for 75C conductors and _not_ for 60C conductors, then I am forced to assume that the breaker can tolerate no more than 75C, and in addition could damage conductors rated only 60C.
-Jon