Possible definitions for the relative momentum of identical particles are considered.PACS numbers: 03.65.TaThe mantra of quantum mechanics is that an observable is a self-adjoint operator and its eigenvalues are the possible results of experiments. This criterion is neither necessary nor sufficient. That it is not sufficient is manifest in the fourth chapter of Gottfried's 1966 book [1] where he points out that perfectly good operators for what are now called Schrödinger cats or grotesque states [2] are not observable. That it is not necessary one learns from that most precisely measured of physical quantities, time, which resists definition as a self-adjoint operator [3,4].In this article we find that another meaningful physical quantity, the relative momentum of identical particles, represents a further failure of the general framework. Momentum is already known to be problematic in two cases: the (single) hard wall [5] and radial momentum [6]. However, an infinite wall is an idealization and the difficulties in the second case could be attributed to the choice of coordinates. With Cartesian coordinates there is no problem.But the relative momentum of a pair of identical particles is unavoidably fundamental. The concept is not straightforward either physically or mathematically. If you cannot know which is which (more precisely, it is simply not defined), how can you attribute a vector to the difference. Mathematically, in the conventional way of dealing with identical particles one assigns identities (say #1 and #2), but works either on the space of symmetric or skewsymmetric states. However, the operator p 1 − p 2 (relative momentum) applied to a state takes you from one space to the other, and is thus not defined in the relevant Hilbert space. On the other hand, there is nothing wrong with the square of the relative momentum, nor with the squares in the examples of the previous paragraph.Physically, however, (p 1 − p 2 ) 2 is not enough. Consider an experiment on a pair of electrons. One can measure momentum with two pairs of position detectors, A and B. They are located so that it is overwhelmingly likely that energy constraints imply that the deduced p A and p B correspond to the individual electrons. The center of mass momentum, P ≡ p