I want to argue that there’s something very basic about the structure of the physical world that’s taken for granted everywhere in physics, but isn’t actually described in any theory. The argument goes like this – What does it take for any physical parameter to be observable? Take the mass of a particle – there are several ways to measure that. For example, if we know its charge, we can watch how the particle’s path gets deflected as it moves through a magnetic field. To do that we have to measure its charge and observe its position at certain times, so we have to be able to measure distances and time-intervals, as well as the strength and orientation of the field. Very generally, every way that any physical parameter can be measured will always involve the measurement of several other parameters. There is no such thing as a simple isolated measurement. So the question is – in what kind of system is any kind of measurement possible? What I want to get at here has nothing to do with whether a human being is involved. It’s a question of what’s needed in the physical situation itself so that the value of a certain parameter can be meaningfully defined, in that situation. The mass of the particle can’t be meaningfully defined except by reference to other parameters, whose values must also be meaningfully defined. Now in the world we live in, there are obviously many observable parameters. For every one of them there are certain interaction-contexts through which they can be measured, all of which involve the measurement of other parameters, in other interaction-contexts. It’s not important to this argument exactly what a “measurement” or “observation” is. It’s enough that we know they’re possible... and that for anything to be in any way observable, there have to be other things that are also observable, in terms of still other things that are also observable. This isn’t an infinite regress – I assume there are only a finite number of basic physical parameters defining each other. But there has to be a certain closure or completeness to the structure of the observable world, as a self-defining system. It seems to me that this inter-referential completeness is a very remarkable characteristic of our universe, and one that no theory I know of takes into account. We tend to take it for granted that if something is real, then of course there will be some way to measure it. After all, if a particle has a certain property X that can’t be observed in any way, then it makes no difference to anything whether that property exists or not – it’s simply meaningless. That makes sense, but I don’t think it undercuts my argument. It’s easy to see that not just any physical system supports the measurement of its own parameters. Imagine a Minimal Newtonian Cosmos consisting of simple point-particles scattered through Euclidean space and time. Say each particle has a certain mass, and no other characteristics, and that they interact with each other only through Newton’s gravity. This seems like a mathematically well-defined system – but there’s no way actually to measure the distance between two particles, or how they move, or what their respective masses are. Interaction in this system may be lawful, but it doesn’t communicate any information about anything, since none of the parameters of the law are defined by the system itself. My point is that in physics, we don’t require that our theoretical models define all their own parameters in terms of each other. We try not to introduce parameters that are – in our universe – unobservable. But we don’t try to model what it is about the structure of our universe that lets anything be observable. I think this is probably why we haven’t come to any clear understanding about Relativity and Quantum theory, and the relationship between them. These theories both (in different ways) make measurement central, but we’re still building models that don’t consider what it takes to make any measurement possible.