As one who believes that the origins of fundamental scientific advances are sometimes experimental, other times theoretical, and often a combination of the two, I do not take sides in the debate between Lincoln Wolfenstein and Harry J. Lipkin in the January 2001 issue of Physics Today (page 13). However, I feel obliged to correct an important error.
Wolfenstein states, “The most exciting results immediately following World War II were the precision atomic experiments verifying the renormalized quantum electrodynamics [QED] of Richard Feynman and Julian Schwinger.” This statement reverses the historical order, since the critical experiments preceded the theoretical explanations. Chronologically, the first critical experiment was the discovery by John Nafe, Edward Nelson, and I. I. Rabi that the hyperfine separation of atomic hydrogen was different from that predicted by the Dirac theory of the electron. 1 This was independently confirmed a few months later. The other critical experiment was the discovery by Willis Lamb and Robert Retherford that the Dirac theory prediction of degeneracy for the 22S1/2 and 22P1/2 atomic hyperfine structure levels in atomic hydrogen was wrong by many times the experimental error of their new measurements. These experiments preceded their theoretical explanations, as I can show.
The Nafe, Nelson, and Rabi letter 1 was received by the Physical Review editor on 19 May 1947, and the first Lamb and Retherford experimental paper 2 was received on 18 June 1947. Those experiments were intensively discussed at the famous Shelter Island theoretical conference held in June 1947, with most attention being devoted to the Lamb-Retherford experiment, which gave a much bigger anomaly and was more easily explained. Hendrick Kramers and others argued that radiative corrections to the energy of an electron in a Coulomb field could produce a shift in the energy levels of hydrogen, and that it might be possible to calculate such an effect approximately, despite the infinities that plagued most radiative correction calculations. Hans Bethe, in an article received on 27 June 1947, 3 published a nonrelativistic field theory calculation that accounted for most of the observed Lamb shift. The first theoretical publication to discuss the experimental hyperfine structure anomaly 4 was received from Gregory Breit on 29 September 1947. However, Breit merely suggested that the experiment was consistent with the electron’s having a magnetic moment different from the Dirac value; he made no effort to provide a QED explanation. Schwinger’s great renormalized field theory, 5 which accounted for both the hyperfine and fine structure anomalies, was received 30 December 1947, seven months after the experiments were reported.
Surely the Nafe, Nelson, and Rabi hyperfine structure experiment was not invented to verify the renormalized electrodynamics, since it was invented by Rabi and me in late 1944. We gave it highest priority because we and others had measured the proton magnetic moment, the electron magnetic moment was given by the Dirac theory, and the electron wavefunction in hydrogen was known. Therefore, the hyperfine separation could be calculated, and the experiment would provide a fundamental test of the Dirac theory. We never mentioned QED, with or without renormalization.
My second experience relating to the sequence of experiment and theory occurred in the summer of 1947 when I visited Cambridge, Massachusetts. Schwinger invited me to lunch and asked me searching questions about the reliability of the experimental hyperfine anomaly. He said he thought he could explain it, but would have to develop a relativistic QED; he was worried about doing all that work if the hyperfine anomaly wasn’t real. I told him I was convinced it was real. He then worked vigorously on this problem, and the following December submitted his paper that correctly accounted for both hyperfine and fine structure experimental anomalies as being due to radiative corrections that could be calculated in his theory, using both mass and charge renormalization.