Voltage is relative. That's all there is to say!
Perhaps it would be more illuminating to think in terms of electric field, which produces the force on the electroscope. Electric field is the spacial derivative (gradient) of voltage, so it has no absolute reference, the constant term disappears. Likewise, voltage can be measured as the integral (along a given path) of electric field; the absolute voltage is the "plus a constant" of integration -- a free variable that doesn't matter to the problem, internally.
Tim
Sorry for bumping an old thread. I came across this thread while trying to find the answer to the same problem on the Web. After getting confused by it for several years, I think I have finally found the answer.
In engineering textbooks, voltage is always defined with respect to a reference, so engineering students believe all voltages are relative. Meanwhile, many physics textbooks on electrostatics say that an object's absolute electric potential is defined with respect to point at infinity for several reasons. So physics students believe an absolute voltage exists in some forms, at least in theory. In fact, it can be really difficult to analyze non-circuit problems in physics if you don't accept the existence of an absolute potential:
1. Imagine a single electron in a vacuum, the absolute electric potential at point A is the work required (with a negative sign) to overcome the electric force to bring a test charge from point A to a point at infinity without any electric field.
2. All objects are formed by subatomic particles, some particles like electrons and protons, carries an elementary charge, and it can be both polarities. Theoretically one knows the exact charge distribution of an object, it's possible to calculate its electric potential and electric field, with respect to point at infinity.
3. Like charges always repel, opposite charges always attract, and both charges always attract neutral objects, it's just Coulomb's law. Even two objects that have never met each other before, still apply electrostatic forces to each other - circuit theory doesn't allow us to analyze this situation but electrostatics can, with respect to point at infinity. Also, if you add charges to all objects to increase their potentials, the outcome of electrostatic experiments may change. This suggest one can detect charges absolutely in theory.
Even in engineering, absolute electric potential is sometimes used when there's no well-defined reference. For example the Human-Body Model of ESD said the human body's capacitance is a "free-space capacitance". In other words, the "self-capacitance" in electrostatics - it's the extra charges needed to increase the absolute electric potential of an object by 1 V. So accepting absolute potential gives you a wildcard when your reference is not well-defined - with respect to what? Just everything else, whatever it may be...
So is voltage absolute or relative, for real? It turns out that, after reading more physics... Ultimately, it's still relative. The trick is to consider what happens when you increase the electric potential of the vacuum itself together with all objects inside it, once you do that, the outcome of all electrostatic experiments remain unchanged. Coulomb's law works just fine. The electric field remains unchanged since the (static) electric field is the gradient of electric potential, and only E causes physical effects, so only the potential difference matters. Furthermore, it turns out that the entire theory of classical electromagnetism is known as gauge-invariant in theoretical physics. Adding a gauge transformation to the electric potential V or the magnetic vector potential A, does not change the observable electric field E and magnetic field B.
So TL;DR: Absolute electric potential exists, in a sense, using a far vacuum as the reference. Sometimes the concept is even useful too. But this definition assumes the electric potential of the "background" vacuum itself is 0 V - which is only a convention. One is free to choose other values.
To put it in another way, one can imagine the universe as a Faraday cage with Perfect Electric Conductor as its boundary condition. It's natural to define the cage itself to be at 0 V for simplicity, but in fact any voltage would do, since the potential of everything inside the cage increases by the same amount so it has no detectable physical effect - observers outside that cage will have different ideas about its electric potential as well.
Now here's the question I don't know yet... Is it possible to approximately measure the "absolute electric potential" defined by electrostatics experimentally? If not a point at infinity, at least a point at somewhere else in the solar system...