Let's all try a little thought experiment. First:
Basic physics problem. You have a wall, a horizontal surface, and a vertical drop. A string is anchored to the wall by means of a tension gauge, which measures tension. The string goes horizontally over the horizontal surface, goes over a pully at the vertical drop, and then goes straight down, with an object that ways X newtons on the end of it. What tension does the spring gauge read?
Answer: For a string, tension is the same on all parts of the string, therefore the gauge should read X newtons.
Now, from here, how can we modify this to be more relevant? BTW, I think the above is essentially Greg's argument...correct me if I'm wrong.
Now, the phenomenon of certain guitars being easier or harder to bend on IS true. Any physical model is either FALSE or INCOMPLETE if it fails to describe what IS KNOWN TO BE TRUE. In other words, no matter how good Greg's argument is, if it doesn't conform to the real world, it's not correct.
So...the question is....what are we missing here?
Consider: what if a string could be approximated by a spring?
IE, the more it is disturbed from its equilibrium length, the more force it will exert.
Now in this analysis there are two forces we are primarily concerned with:
1) The force that we exert on the string (tension, to bring it up to pitch).
2) The force the string exerts in order to get back to its equilibrium position.
My guess is that 2) is what's responsible for the perception of more or less bendability. Think about it, the string wants to be STRAIGHT and at a certain natural equilibrium length. If you put an angle into the headstock, the string is in a non-equilibrium position and will thus exert force (in this case on the nut).
In the case of an angled headstock, fretting or bending will put the string FURTHER from its equilibrium length. And from high school physics, the force the string exerts is directly proportional to the displacement from equilibrium length.
Shoot me down if I missed something.
Cheers
Yike