Transfer the state, using a particle already there
In the ideal protocol, Alice receives an unknown input qubit and shares an entangled pair with Bob. A joint measurement on Alice’s two systems produces a classical label. Bob uses that label to choose a correction on his half of the pair. The resulting state can match the input without sending the original input particle to him.
The protocol consumes shared entanglement and the original input is no longer available as an independent perfect copy. Knowing the classical label does not disclose the unknown amplitudes. Bennett and colleagues’ 1993 proposal made these resource requirements explicit.
Without the label, no readable message
For an ideal maximally entangled qubit pair and a complete Bell measurement, the four outcomes correspond to four known Pauli corrections. The expression below averages over these outcomes before Bob receives the label. The input density operator has unit trace, and the identity acts on Bob’s two-dimensional qubit.
His unconditioned state is independent of the input. Sorting data after the classical message can reveal the state transfer, but it cannot make that message arrive earlier. The two classical bits of a complete ideal protocol and the selected successful branch of an optical experiment are different operating contracts.
The 1997 optical experiment selected one branch
Bouwmeester and colleagues combined an input photon with one photon from an entangled pair at a beam splitter. Suitable joint detections selected an antisymmetric Bell-state branch. The receiver’s photon was then analyzed under that condition, using several input polarizations.
This was a probabilistic optical realization. It did not distinguish and correct all four Bell outcomes on every trial. Testing superposition inputs matters because a device that merely recognizes one preferred polarization basis can imitate successful transfer on that basis while failing on other states.
Read the evidence: D. Bouwmeester et al. (1997), Experimental Quantum Teleportation, Nature 390, 575–579 — ETH-hosted copy
Background subtraction and heralding are different records
The paper distinguishes threefold coincidence curves with a measured spurious contribution subtracted from fourfold records conditioned on detection of the input photon’s partner. The latter were shown without that background subtraction for two nonorthogonal inputs. These are different analyses with different event selections.
A visibility is not automatically an unconditional success probability or a process fidelity. To interpret a reported value, identify the denominator, losses, subtraction and conditioning first. In particular, verification that also detects the output photon should not be described as an output left available for another task on every herald.
Read the evidence: D. Bouwmeester et al. (1997), Experimental Quantum Teleportation, Nature 390, 575–579 — ETH-hosted copy
Swapping changes who shares entanglement
Pan and colleagues’ 1998 experiment began with two entangled photon pairs. A Bell-state measurement on one member of each pair conditionally prepared entanglement between the two remaining photons. Those two need not have directly interacted with one another.
The statement concerns a conditioned joint state and its verification. It does not mean each remaining photon carries a locally readable announcement, or that a later data selection sends a signal into the past. Bell-inequality tests and teleportation benchmarks also ask different statistical questions.
Read the evidence: J.-W. Pan et al. (1998), Experimental Entanglement Swapping: Entangling Photons That Never Interacted, Phys. Rev. Lett. 80, 3891 — author abstract
Compare a protocol with the experiment that tests it
A teaching circuit can supply an entangled pair, choose a Bell outcome, apply a correction and display an output state exactly. That is useful for understanding the ideal channel. It does not establish real source brightness, indistinguishability, detector efficiency or network operation.
Use the linked teleportation and tomography Labs for their declared models, then return to the paper’s event definition. A convincing evidence claim names the input ensemble, accepted branch, output test and unsuccessful trials. This keeps the quantum-information achievement precise without turning a conditional demonstration into a universal transport service.
Read the evidence: C. H. Bennett et al. (1993), Teleporting an Unknown Quantum State via Dual Classical and Einstein-Podolsky-Rosen Channels, Phys. Rev. Lett. 70, 1895–1899 — university-hosted copyD. Bouwmeester et al. (1997), Experimental Quantum Teleportation, Nature 390, 575–579 — ETH-hosted copy