Heisenberg's uncertainty principle was originally posed for the limit of the accuracy of simultaneous measurement of non-commuting observables as stating that canonically conjugate observables can be measured simultaneously only with the constraint that the product of their mean errors should be no less than a limit set by Planck's constant. However, Heisenberg with the subsequent completion by Kennard has long been credited only with a constraint for state preparation represented by the product of the standard deviations. Here, we show that Heisenberg actually proved the constraint for the accuracy of simultaneous measurement but assuming an obsolete postulate for quantum mechanics. This assumption, known as the repeatability hypothesis, formulated explicitly by von Neumann and Schrödinger, was broadly accepted until the 1970s, but abandoned in the 1980s, when completely general quantum measurement theory was established. We also survey the author's recent proposal for a universally valid reformulation of Heisenberg's uncertainty principle under the most general assumption on quantum measurement.
In two articles, the authors claim that the Heisenberg uncertainty principle limits the precision of simultaneous measurements of the position and velocity of a particle and refer to experimental evidence that supports their claim. It is true that ever since the inception of quantum mechanics, the uncertainty relation that corresponds to a pair of observables represented by non-commuting operators is interpreted by many scientists and engineers, including Heisenberg himself, as a limitation on the accuracy with which observables can be measured. However, such a limitation cannot be deduced from the postulates and theorems of quantum thermodynamics.
The uncertainty principle is commonly regarded as a fundamental limitation ofquantum mechanics, often attributed directly to Heisenberg’s 1927 analysis.In this work, we revisit the original formulation of the principle and carefullydistinguish Heisenberg’s measurement–disturbance argument from the latermathematical uncertainty relations derived by Kennard, Robertson, andSchrödinger. We show that the standard variance-based formulation relies onthe prior introduction of fluctuation operators and ensemble averages, therebypresupposing statistical indeterminacy rather than deriving it from firstprinciples. From a physical perspective, Heisenberg’s argument is shown toreflect constraints imposed by invasive measurement techniques rather than auniversal prohibition on simultaneous knowledge of conjugate variables.Motivated by this distinction, we propose a conjecture that, in the absence ofmeasurement-induced energy transfer, the simultaneous determination ofposition and momentum may, in principle, be achievable. Our analysis suggeststhat quantum uncertainty, as traditionally formulated, encodes limitations ofcurrent observational methods rather than an intrinsic property of nature. \section{Heisenberg’s original observation and formulation (1927)} In his seminal 1927 paper \emph{Über den anschaulichen Inhalt der quantentheoretischen Kinematik und Mechanik} \cite{Heisenberg1927}, Werner Heisenberg sought to clarify the physical meaning of quantum mechanics rather than to establish a rigorous mathematical theorem. His analysis was based on thought experiments concerning the act of measurement, most notably the
γ-ray microscope. Heisenberg argued that any attempt to localize a particle within a spatial region
Δx requires radiation of wavelength comparable to
Δx, which inevitably transfers an uncontrollable momentum
Δp to the particle. As a consequence of this measurement-induced disturbance, he proposed that the product of the inaccuracies in position and momentum measurements satisfies a relation of the order\begin{equation}\Delta x\, \Delta p \sim h .\end{equation}Here,
Δx and
Δp denote operational measurement imprecisions arising from the interaction between the measuring apparatus and the system, rather than statistical variances of quantum observables. Heisenberg did not introduce operators, Hilbert space, or probability distributions, nor did he provide a formal inequality with a precise lower bound. The argument was explicitly heuristic and qualitative, emphasizing limitations imposed by the measurement process itself. Consequently, the original uncertainty principle was formulated as a statement about measurement disturbance and experimental resolution, not as a universal mathematical constraint on quantum states. \section{Standard mathematical proof of the position--momentum uncertainty relation} \subsection{Setup and definitions} Let
H be a Hilbert space and let
∣ψ⟩∈H be anormalized state,
⟨ψ∣ψ⟩=1. Let
A and
B beself-adjoint operators (observables) on a common dense domain containing
∣ψ⟩, such that all expressions below are well-defined. Define expectation values\cite{Kennard1927,Robertson1929,Schrodinger1930,Ballentine1998,Sakurai1994}\begin{equation}\langle A \rangle := \langle \psi | A | \psi \rangle,\qquad\langle B \rangle := \langle \psi | B | \psi \rangle .\end{equation}Define the fluctuation (centered) operators\begin{equation}\Delta A := A - \langle A \rangle, \qquad \Delta B := B - \langle B \rangle .\end{equation}The variances are\begin{equation}\sigma_A^2 := \langle (\Delta A)^2 \rangle = \langle \psi | (\Delta A)^2 | \psi \rangle,\qquad\sigma_B^2 := \langle (\Delta B)^2 \rangle = \langle \psi | (\Delta B)^2 | \psi \rangle.\end{equation} \subsection{Key inequality: Cauchy--Schwarz in Hilbert space} Define vectors\begin{equation}|\phi\rangle := \Delta A |\psi\rangle, \qquad |\chi\rangle := \Delta B |\psi\rangle .\end{equation}By the Cauchy--Schwarz inequality,\begin{equation}\langle \phi|\phi\rangle \, \langle \chi|\chi\rangle \ge |\langle \phi|\chi\rangle|^2 .\end{equation}Compute each factor:\begin{equation}\langle \phi|\phi\rangle= \langle \psi| (\Delta A)^\dagger (\Delta A) |\psi\rangle= \langle \psi| (\Delta A)^2 |\psi\rangle= \sigma_A^2,\end{equation}and similarly
⟨χ∣χ⟩=σB2. Therefore\begin{equation}\sigma_A^2 \sigma_B^2 \ge \left| \langle \psi| \Delta A \Delta B |\psi\rangle \right|^2= |\langle \Delta A \Delta B \rangle|^2 .\label{eq:cs_step}\end{equation} \subsection{Robertson--Schr\"odinger decomposition} Decompose
ΔAΔB into commutator and anticommutator parts:\begin{equation}\Delta A \Delta B=\frac{1}{2}\bigl[\Delta A,\Delta B\bigr]+\frac{1}{2}\bigl\{\Delta A,\Delta B\bigr\},\end{equation}where
[X,Y]=XY−YX and
{X,Y}=XY+YX. Taking expectation values gives\begin{equation}\langle \Delta A \Delta B \rangle=\frac{1}{2}\langle [\Delta A,\Delta B]\rangle+\frac{1}{2}\langle \{\Delta A,\Delta B\}\rangle .\label{eq:decomp_exp}\end{equation} Now note the following facts for self-adjoint
A,B:\begin{itemize}\item The operator
[ΔA,ΔB] is anti-self-adjoint, hence
⟨[ΔA,ΔB]⟩ is purely imaginary.\item The operator
{ΔA,ΔB} is self-adjoint, hence
⟨{ΔA,ΔB}⟩ is real.\end{itemize}Therefore the two terms in \eqref{eq:decomp_exp} are orthogonal in the complex plane(real vs.\ imaginary), and we may write\begin{equation}|\langle \Delta A \Delta B \rangle|^2=\left(\frac{1}{2}|\langle [\Delta A,\Delta B]\rangle|\right)^2+\left(\frac{1}{2}\langle \{\Delta A,\Delta B\}\rangle\right)^2 .\label{eq:RS_identity}\end{equation} Substituting \eqref{eq:RS_identity} into \eqref{eq:cs_step} yields theSchr\"odinger uncertainty relation:\begin{equation}\boxed{\sigma_A^2 \sigma_B^2 \ge\left(\frac{1}{2}|\langle [\Delta A,\Delta B]\rangle|\right)^2+\left(\frac{1}{2}\langle \{\Delta A,\Delta B\}\rangle\right)^2}.\label{eq:schrodinger}\end{equation} Since constants commute with everything,\begin{equation}[\Delta A,\Delta B] = [A-\langle A\rangle,\; B-\langle B\rangle] = [A,B].\end{equation}Hence \eqref{eq:schrodinger} can be written as\begin{equation}\boxed{\sigma_A^2 \sigma_B^2 \ge\left(\frac{1}{2}|\langle [A,B]\rangle|\right)^2+\left(\frac{1}{2}\langle \{\Delta A,\Delta B\}\rangle\right)^2}.\end{equation} \subsection{Robertson inequality (weaker but widely quoted)} Dropping the nonnegative anticommutator term gives Robertson's inequality:\begin{equation}\boxed{\sigma_A \sigma_B \ge \frac{1}{2}|\langle [A,B]\rangle|}.\label{eq:robertson}\end{equation} \subsection{Specialization to position and momentum} Take
A=x and
B=p (one spatial dimension), with canonical commutation relation\begin{equation}[x,p] = i\hbar.\end{equation}Therefore,\begin{equation}|\langle [x,p]\rangle|= |\langle i\hbar \rangle|= \hbar,\end{equation}and \eqref{eq:robertson} yields the standard position--momentum uncertainty relation\begin{equation}\boxed{\sigma_x \sigma_p \ge \frac{\hbar}{2}.}\end{equation} \subsection{Condition for equality (minimum-uncertainty states)} Equality in Cauchy--Schwarz holds iff the vectors are linearly dependent:\begin{equation}\Delta A|\psi\rangle = \lambda \, \Delta B|\psi\rangle\quad \text{for some complex } \lambda.\end{equation}For
A=x and
B=p, this condition leads (in the
x-representation) to afirst-order differential equation whose normalized solutions are Gaussianwave packets. Hence Gaussian states saturate
σxσp=ℏ/2. \subsection{Definition of fluctuation operators and their role in the uncertainty relation} In the mathematical formulation of quantum uncertainty developed after Heisenberg’s original 1927 heuristic proposal, the quantities entering the uncertainty relation were redefined in statistical terms. In particular, in his 1929 paper, Robertson introduced the operators. \noindent\textbf{Critical observation.}The introduction of the fluctuation operators \begin{center}\fcolorbox{red}{white}{
ΔA=A−⟨A⟩,ΔB=B−⟨B⟩}\end{center} \textbf{constitutes a nontrivial mathematicalassumption rather than a derived physical necessity. In the Robertson--Schrödinger formalism, the uncertainty of observables is defined \emph{a priori} in terms ofstatistical deviations from expectation values, thereby presupposing the verynotion of indeterminacy that the uncertainty principle is intended to justify.In other words, the operators
ΔA and
ΔB encode uncertainty byconstruction, since their variances measure ensemble spread relative to a mean.The subsequent derivation of a lower bound on
σAσB thereforeestablishes a constraint on these predefined statistical quantities, rather than demonstrating that physical observables must be uncertain from first principles.Consequently, the uncertainty relation obtained in this manner is logicallyconditional upon the initial choice of variance-based definitions and does not,by itself, prove the existence of intrinsic measurement indeterminacy or disturbance in individual quantum events.} \textit{They actully assumed and defined the spread and uncertainity initial in both the quantities and proved the uncertainity principle.} where
A and
B are self-adjoint operators representing observables and
⟨A⟩=⟨ψ∣A∣ψ⟩ denotes the expectationvalue in a normalized quantum state
∣ψ⟩ \cite{Robertson1929}.These operators describe deviations of observables from their mean valuesand are introduced in direct analogy with classical statistical fluctuations.The corresponding variances,
σA2=⟨(ΔA)2⟩ and
σB2=⟨(ΔB)2⟩,measure the statistical spread of measurement outcomes over an ensembleof identically prepared systems
The uncertainty relation formulated by Heisenberg in 1927 describes a trade-off between the error of a measurement of one observable and the disturbance caused on another complementary observable so that their product should be no less than a limit set by Planck's constant. In 1980, Braginsky, Vorontsov, and Thorne claimed that this relation leads to a sensitivity limit for gravitational wave detectors. However, in 1988 a model of position measurement was constructed that breaks both this limit and Heisenberg's relation. Here, we discuss the problems as to how we reformulate Heisenberg's relation to be universally valid and how we experimentally quantify the error and the disturbance to refute the old relation and to confirm the new relation.
Does the Heisenberg uncertainty principle refer to errors of measurement, or to the spread of values of the physical variables intrinsic to a particle’s state, or to some combination of these? The most commonly quoted form of the uncertainty principle relates the spread of an ensemble of separate measurements of some variable to the analogous spread of its conjugate variable. In contrast, Heisenberg’s original argument for the uncertainty principle involved the perturbation to a particle’s state by a measurement of one variable, which affects one’s ability to predict the outcome of a subsequent measurement of the conjugate variable. The relation between these two views of the uncertainty principle is discussed in this paper. A familiar example is considered: an ensemble of identically prepared particles passing through a slit, and after further propagation being detected. From this arrangement it is possible to infer joint (although necessarily imprecise) information on both transverse position and conjugate momentum for each member of the ensemble. It is shown that in this case of joint measurement the product of standard deviations for the measurement outcomes is at least twice as large as the lower bound implied by the usual uncertainty principle. The discussion is meant to help clarify the different roles played in the various statements of the uncertainty principle by the initial state and by measurement error.