Observables Before Operators

In modern quantum mechanics, an observable is represented by a self-adjoint operator, with its possible outcomes organized by the operator’s spectrum. Historically, the order of discovery was less tidy. Physicists first had measured spectral lines, transition frequencies, intensities, and selection patterns. The operator concept grew out of the attempt to organize those data without pretending that electrons followed observable classical orbits inside atoms.

This page is the historical bridge. The formal definitions of Observables, Operators, and Spectral Decomposition belong in Core Formalism. Here the question is why measurable quantities began to look like arrays, matrices, and finally operators.

Spectral Lines as Observable Data

Before quantum mechanics had a Hilbert-space formalism, spectroscopy already supplied precise facts. A gas discharge, flame, or absorption cell did not produce arbitrary frequencies. It produced characteristic lines with reproducible positions, finite widths, relative strengths, polarizations, and selection patterns.

The line positions were the most direct clue. In modern notation, a transition from a higher atomic energy $E_m$ to a lower energy $E_n$ emits radiation with

h\nu_{mn} = E_m-E_n, \qquad \omega_{mn} = \frac{E_m-E_n}{\hbar}.

This formula is now so familiar that it can hide the historical point. The observed object was not an electron path. It was a line in a spectrum, indexed by two stationary levels. The pair of labels $m,n$ mattered from the beginning.

Line intensities added a second layer. Two transitions can have different strengths even when their frequencies are both allowed by energy differences. A theory of frequencies alone is therefore incomplete. The emerging theory needed quantities that could say not just where a line appears, but how strongly a transition couples to radiation.

Transition Quantities

Old quantum theory kept much of the language of classical orbits. For a periodic classical motion, a coordinate could be expanded into Fourier components,

x(t) = \sum_\alpha x_\alpha e^{i\alpha\omega t}.

The correspondence principle suggested that high quantum numbers should reproduce classical Fourier behavior in an appropriate limit. But the actual quantum data were transition data. Heisenberg’s reinterpretation replaced a classical harmonic label $\alpha$ by a pair of quantum labels:

x_\alpha \quad\leadsto\quad X_{mn}, \qquad \alpha\omega \quad\leadsto\quad \omega_{mn}.

The quantity $X_{mn}$ was not a measured position at an instant. It was a transition quantity. In a later, modern reconstruction one writes

X_{mn} = \langle m\vert X\vert n\rangle,

but bra-ket notation and the abstract operator meaning came later. The historically decisive step was already present: the useful quantity carried two state labels, and it was tied to observed transition behavior.

A schematic connection to line strength is

\text{line strength} \propto \lvert X_{mn}\rvert^2

in simple electric-dipole settings, with details depending on polarization, degeneracy, angular momentum, and the interaction Hamiltonian. This should not be read as a complete radiation theory. It is the basic reason transition entries, not hidden orbits, became central.

Matrix Entries

Once quantities are indexed by pairs of states, they naturally assemble into arrays:

X \sim \begin{pmatrix} X_{11} & X_{12} & X_{13} & \cdots \\ X_{21} & X_{22} & X_{23} & \cdots \\ X_{31} & X_{32} & X_{33} & \cdots \\ \vdots & \vdots & \vdots & \ddots \end{pmatrix}.

Diagonal entries are associated with a stationary state. Off-diagonal entries connect different states and are therefore the natural place for transition information. This is why the matrix form was not a decorative linear-algebra choice. It matched the empirical grammar of spectroscopy.

The multiplication rule also came from transition structure. If a product of two quantities passes through an intermediate state $k$ , the frequencies compose:

\omega_{mk} + \omega_{kn} = \omega_{mn}.

The associated product of arrays is

(XY)_{mn} = \sum_k X_{mk}Y_{kn}.

This is matrix multiplication. Reversing the order gives

(YX)_{mn} = \sum_k Y_{mk}X_{kn},

which need not agree with $(XY)_{mn}$ . Noncommutativity entered because the transition quantities had a composition law unlike ordinary multiplication of classical numbers.

Operators as Modern Formalization

The modern operator concept gives a representation-independent version of this story. Choose an energy basis $\{\lvert n\rangle\}$ . A quantum observable $A$ has matrix elements

A_{mn} = \langle m\vert A\vert n\rangle.

The matrix depends on the chosen basis. The operator $A$ is the object behind all its representations. Dirac’s transformation theory made this basis flexibility explicit, and later Hilbert-space formalism made it mathematically systematic.

For a finite-dimensional sharp observable with discrete spectrum, modern spectral decomposition writes

A = \sum_a aP_a,

where the eigenvalues $a$ are possible ideal measurement outcomes and the projectors $P_a$ determine probabilities. In a nondegenerate basis,

P_a = \lvert a\rangle\langle a\rvert.

This modern formula packages two historical lessons:

measurable quantities are organized by possible outcomes;
transitions and representation changes require matrix elements between states.

The operator language therefore did not replace experimental observables with abstract symbols. It gave a durable mathematical form to the spectral and transition data that had forced the break with classical mechanics.

What Changed Conceptually

Classically, a quantity such as position is usually treated as a value along a trajectory. In early atomic physics that picture became increasingly strained. Spectroscopy gave pairs of levels, transition frequencies, intensities, and selection rules, but not observable electron orbits.

The conceptual change was this:

a measurable quantity need not be a pre-existing classical value revealed by looking;
a quantum quantity can have diagonal information about stationary states and off-diagonal information about transitions;
products of quantities can depend on order;
the same physical quantity can have different matrix representations in different bases.

This is why “observable” in quantum mechanics is not just a synonym for “what one can see.” It is a structured mathematical object whose spectra, matrix elements, and algebra determine observable consequences.

What This Page Does Not Claim

This history should not be turned into a slogan that only directly observable quantities are meaningful. Quantum theory uses states, amplitudes, phases, Hamiltonians, fields, and idealizations that are not themselves read off a dial. The better lesson is narrower: early matrix mechanics learned to avoid untestable classical orbital pictures when the spectral data themselves suggested a new algebraic structure.

It is also too simple to say that every laboratory measurement is a self-adjoint operator. The standard operator picture is the canonical first language for sharp observables, but realistic measurements may require projectors, POVMs, detector models, and open-system dynamics. Those refinements belong in the measurement volumes.

Common Mistakes

Thinking “observable” historically meant only a quantity visible to the eye. Spectral frequencies, intensities, and transition patterns were observable data even though the underlying atomic processes were inferred.
Treating an operator as if it were merely a table of possible values. Off-diagonal matrix elements carry transition and representation information.
Assuming line frequencies alone determine line intensities. Energies set allowed frequencies; matrix elements and selection rules control strengths.
Reading modern Hilbert-space language back into 1925 without warning. It is useful as reconstruction, but it was not the starting notation.
Equating observability with philosophical positivism. The historical issue was the failure of classical orbit pictures, not a ban on theoretical entities.
Forgetting that a matrix is basis-dependent, while the modern operator is the representation-independent object.

Cross-Links

References

W. Heisenberg, “Über quantentheoretische Umdeutung kinematischer und mechanischer Beziehungen,” Zeitschrift für Physik 33, 879-893, 1925, DOI: 10.1007/BF01328377.
M. Born and P. Jordan, “Zur Quantenmechanik,” Zeitschrift für Physik 34, 858-888, 1925, DOI: 10.1007/BF01328531.
M. Born, W. Heisenberg, and P. Jordan, “Zur Quantenmechanik II,” Zeitschrift für Physik 35, 557-615, 1926, DOI: 10.1007/BF01379806.
P. A. M. Dirac, “The fundamental equations of quantum mechanics,” Proceedings of the Royal Society A 109, 642-653, 1925, DOI: 10.1098/rspa.1925.0150.
J. R. Rydberg, “Recherches sur la constitution des spectres d’emission des elements chimiques,” Kungliga Svenska Vetenskapsakademiens Handlingar 23, 1-177, 1890.
B. L. van der Waerden, ed., Sources of Quantum Mechanics, Dover, 1968.
M. Jammer, The Conceptual Development of Quantum Mechanics, 2nd ed., American Institute of Physics, 1989.
J. Mehra and H. Rechenberg, The Historical Development of Quantum Theory, Springer, 1982-2001.

Exercises

Explain why a spectral line naturally carries two state labels in the modern reconstruction.

Solution

A line frequency corresponds to an energy difference:

h\nu_{mn} = E_m-E_n.

The line is therefore associated with a transition between two stationary states, not with one isolated state label. This pair-label structure is what later appears as a matrix element such as $\langle m\vert A\vert n\rangle$ .

Why do line intensities require more than the Bohr frequency condition?

Solution

The Bohr frequency condition determines the frequency associated with an energy difference. It does not say how strongly the atom couples to radiation in that transition. In modern language, transition strengths depend on matrix elements, selection rules, degeneracies, and the interaction with the electromagnetic field.

Let $X$ and $Y$ be transition arrays. Explain in words why $(XY)_{mn}$ contains a sum over intermediate labels.

Solution

The product can connect an initial label $n$ to a final label $m$ through an intermediate label $k$ . Since many intermediate states may contribute, the product sums over $k$ :

(XY)_{mn} = \sum_k X_{mk}Y_{kn}.

This is the same structural rule as matrix multiplication.

In the nondegenerate finite-dimensional case, an observable has

A = \sum_a a\lvert a\rangle\langle a\rvert.

What part of this formula gives the possible outcomes, and what part determines the measurement alternatives?

Solution

The numbers $a$ are the possible ideal measurement outcomes. The projectors $\lvert a\rangle\langle a\rvert$ determine the corresponding alternatives and are used to compute probabilities through the Born rule. The observable packages both outcome labels and projectors.