The book states the key physical requirement after constructing the spin operators: observables must be represented by operators whose eigenvalues can be measurement results. Since measurement results are real, the relevant operators must be Hermitian. This page expands that point and connects it to the matrix notation already introduced.
For a matrix, the adjoint is the conjugate transpose:
An operator is Hermitian when it equals its adjoint:
The Pauli matrices \(\hat\sigma_z\) and \(\hat\sigma_x\) satisfy this condition. That is why they are valid candidates for spin observables.
Suppose \(\hat A|a\rangle=a|a\rangle\) and \(\hat A\) is Hermitian. Consider the expectation value in the normalized eigenstate:
Taking the complex conjugate and using \(\hat A^\dagger=\hat A\) gives the same number. Therefore
The eigenvalue is real, as required for a measurement result.
The same logic applies to any normalized state. If \(\hat A\) is Hermitian, then
So expectation values of observables are real even when the state vector has complex components.
For Hermitian operators, eigenvectors with different eigenvalues are orthogonal. This matters physically: different sharp measurement results must be distinguishable. In a Stern-Gerlach apparatus, the upper and lower beams are different channels, represented by orthogonal states.
This orthogonality is what makes it meaningful to treat the outgoing beams as separate alternatives in the Born rule.
The possible results of the dimensionless spin measurement are \(+1\) and \(-1\). The matrices \(\hat\sigma_z\) and \(\hat\sigma_x\) have exactly those real eigenvalues. They therefore satisfy the physical requirement emphasized by the chapter: observables must have real values that an apparatus can record.
The postulate for observables is not simply a naming convention. It says that the measurable content of the theory is encoded in Hermitian operators. Their eigenvalues are the possible measurement results, and their eigenvectors define the states with definite values.
This page is designed to support short guided exercises on: Why observables must have real eigenvalues, and how Hermitian conjugation controls the matrix representation.
Use these anchors to design compact exercises. The exercise should be answerable from this page plus standard algebra, while longer derivations, full worked examples and broader context should point back to the original book.