Fermions and Bosons Are a Three-Dimensional Theorem, Not a Law of Quantum Mechanics — Epoche C2
The claim Undergraduate courses teach that every particle is either a fermion or a boson, and present this as a basic fact about quantum mechanics. It is not. It is a theorem, and the theorem has a premise that is easy to miss: that the particles move in three or more spatial dimensions. Remove that premise and the conclusion fails. This essay sets out the deduction, shows precisely where it stops, and then asks how much of the two-dimensional alternative has actually been measured. Following the condensed-matter convention of the works cited, magnetic quantities are Gaussian — fields in gauss, magnetic length $\ell_B = \sqrt{\hbar c/eB}$, flux quantum $\Phi_0 = hc/e = 4.14\times10^{-7}$ G cm$^2$, Coulomb energy $e^2/\varepsilon\ell_B$ with $e = 4.80\times10^{-10}$ esu — while the transport measurements are quoted in SI, as their sources are, with $e = 1.60\times10^{-19}$ C. The premise, stated exactly Two particles are identical when no measurement distinguishes them. The physical consequence is that the state after swapping them must describe the same physics as the state before, so the many-body wavefunction $\psi$ can change only by a phase: $$\psi(\ldots,\mathbf{r}_i,\ldots,\mathbf{r}_j,\ldots) \;\longrightarrow\; e^{i\theta}\,\psi(\ldots,\mathbf{r}_j,\ldots,\mathbf{r}_i,\ldots),$$ where $\mathbf{r}_i$ and $\mathbf{r}_j$ are the positions of the two particles and $\theta$ is a real number called the exchange phase. Nothing so far restricts $\theta$. The restriction comes from asking what an exchange is . It is not an instantaneous relabelling; it is a continuous path in the space of configurations that ends with the two particles interchanged. Leinaas and Myrheim made this point precisely in 1977: the allowed phases are labels of the closed paths in configuration space that cannot be deformed into one another, which is the fundamental group $\pi_1$ of that space. The deduction in three dimensions Take two particles and use the relative coordinate $\mathbf{r} = \mathbf{r}_1 - \mathbf{r}_2$. The particles cannot sit on top of each other, so $\mathbf{r} \neq 0$: the configuration space is $\mathbb{R}^3$ with the origin removed, and an exchange is a path from $\mathbf{r}$ to $-\mathbf{r}$. Now do the exchange twice. The result is a closed loop that starts and ends at $\mathbf{r}$. In three dimensions this loop can be lifted out of the plane and shrunk to a point, because a single removed point does not obstruct a loop in three-dimensional space — there is always a direction to go around it. A continuously contractible loop must carry no phase. This forces $$e^{2i\theta} = 1 \quad\Longrightarrow\quad \theta = 0 \ \text{or}\ \theta = \pi ,$$ which are the boson and the fermion. The whole dichotomy is that one step. Notice what does the work: it is the contractibility of the double loop, not the Pauli principle, not spin, not relativity. Spin enters only later, through the spin–statistics theorem, which tells you which of the two options a given field must take. Where the deduction stops In two dimensions the same double loop encircles the removed origin, and a loop that winds once around a puncture in a plane cannot be shrunk without crossing it. The loops are now classified by an integer winding number rather than by a two-valued parity: $\pi_1$ is $\mathbb{Z}$ rather than $\mathbb{Z}_2$. The constraint $e^{2i\theta} = 1$ therefore never arises, and $\theta$ may take any value. Wilczek named such particles anyons in 1982. For $N$ particles the same replacement is the braid group in place of the permutation group; when the phases are replaced by matrices acting on a degenerate set of states, the particles are non-Abelian, and the order of two braids matters. It is worth being clear about what "two-dimensional" means physically. It does not mean a thin sample. It means that the excitations concerned have no low-energy motion in the third direction, so that their world lines genuinely live in a plane plus time. That is why quantum Hall systems are the natural home for anyons. The realisation: Laughlin quasiparticles Laughlin's 1983 trial wavefunction for the fractional quantum Hall state at filling factor $\nu = 1/m$ with $m$ odd is $\Psi_m = \prod_{i<j}(z_i - z_j)^m \exp\!\left(-\sum_k |z_k|^2/4\ell_B^2\right)$, with $z_k = x_k + iy_k$ the complex coordinate of the $k$th electron. The filling factor $\nu$ is the number of electrons per flux quantum, so $\nu = n_e \Phi_0 / B$ with $n_e$ the areal electron density. A typical GaAs sheet with $n_e = 1\times10^{11}$ cm$^{-2}$ reaches $\nu = 1/3$ at $B = 3 n_e \Phi_0 = 3 \times 10^{11} \times 4.14\times10^{-7} = 1.24\times10^{5}$ G, that is 12.4 tesla, where $\ell_B = 7.3\times10^{-7}$ cm. The elementary excitation of this state is a vortex: one quantum of flux is expelled, and the charge deficit it leaves is $e^* = e/m$. Attaching a flux $\Phi_0$ to a charge $e/m$ fixes the statistics by the Aharonov–Bohm phase alone. Carrying one composite all the way round another gives $(e^*/\hbar c)\Phi_0 = (e/3)(hc/e)/(\hbar c) = 2\pi/3$ for $m = 3$, and an exchange is half a circuit, so $\theta = \pi/3$. The scale of everything is the Coulomb energy $e^2/\varepsilon \ell_B$; with the GaAs dielectric constant $\varepsilon = 12.9$ this is 15.3 meV, and the measured $\nu = 1/3$ gap is roughly 0.03 of it, about 0.46 meV, or 5.3 K. That is why these experiments run at tens of millikelvin. What has been measured The charge is settled. Shot noise in a weakly pinched constriction obeys the Schottky formula $S_I = 2e^*I_B$, where $S_I$ is the current noise power and $I_B$ the backscattered current; the noise counts the carriers, so it fixes their charge without any model of the state. For $I_B = 1$ nA, $e^* = e/3 = 5.34\times10^{-20}$ C gives $S_I = 1.07\times10^{-28}$ A$^2$ Hz$^{-1}$, exactly one third of the $3.2\times10^{-28}$ expected for electrons. Two groups reported this in 1997, and the factor of three is not subtle. The statistics is the harder claim, and