We tend to treat numbers and measurements as though they belonged directly to things. We say that a structure « is 12 », that a crystal « has a 6-fold symmetry » or that an organism « chooses 5 ». Yet the real never delivers an isolated number. It presents an organisation, and we then decide what we are going to distinguish, count, compare and measure. The number appears at the end of that operation.
This does not mean that numbers are arbitrary or purely invented. We choose the unit, the object counted, the scale and the system of measurement, but we do not freely choose the relations that resist our operations. Once we decide to count the faces of a spherical trivalent network, Euler's formula imposes the balance. Once we require a regular tiling of the plane, the angular equation leaves only three solutions.
Measurement is constructed, but it is constructed under the resistance of the real.
The problem stems from a frequent confusion between three levels. First there is the phenomenon or the real structure. Then there is the operation by which we isolate and measure it. Finally there is the number obtained. When these levels are collapsed into one another, the number ends up being taken for the ultimate property of the thing, or even for its cause. We then confuse numerical regularity with the mechanism that produces it, and measurement with the reality it represents.
Yet one and the same phenomenon can receive several numerical descriptions depending on what is counted. A sphere can lead to twelve pentagons in a trivalent network; to a total topological charge of 2 in a nematic field, where the Poincaré–Hopf theorem requires the sum of the charges to equal the Euler characteristic; to four defects of charge +1/2 that realise this same total; or to a total curvature of 4π by the Gauss–Bonnet theorem. These values do not contradict one another. They are four faces of the same invariant χ = 2, grasped through different objects, units and operations.
The converse is equally true: one and the same value may arise from unrelated mechanisms. The 12 of the fullerenes does not have the same origin as the 12 of the maximum number of identical spheres in contact with a central sphere in three-dimensional space. The first results from a topological balance, the second from a metric packing. Numerical coincidence therefore never suffices to establish an identity of structure or of cause.
This obliges us to revise the way we speak about measurements. A measurement is not a neutral photograph of the real. It is a relation between a phenomenon, an operation, a unit and a theoretical framework. It always answers a precise question. Measuring a length, counting neighbours, calculating an Euler characteristic or identifying an order of rotation does not reveal the same dimension of the object.
Interpreting a number thus requires specifying what was counted, the operation employed, the framework adopted and the constraint that makes the result necessary.
The number is a stabilised trace of the relation established with the real. That trace can be rigorous, universal and demonstrable without constituting the cause of what it describes. The constraints examined in Constraint as the matrix of form show the origin of certain regularities. The present text specifies their status as the results of operations confronted with relations that resist them.
A relation leaves a trace when it meets a support capable of retaining it, which presupposes an asymmetry, an irreversibility or some form of memory. That trace nevertheless underdetermines the relation that produced it. The two occurrences of the number 12 mentioned above arise from distinct mechanisms. Moving back from the number to its cause therefore demands reconstructing the operation and the constraints whose result it preserves.
A number never says on its own what a thing is. It says what a determinate operation allows us to grasp of it.