Physicists have pondered whether the fine-structure constant is in fact constant, or whether its value differs by location and over time. A varying has been proposed as a way of solving problems in
cosmology and
astrophysics.
String theory and other proposals for going beyond the
Standard Model of particle physics have led to theoretical interest in whether the accepted
physical constants (not just ) actually vary. In the experiments below, represents the change in over time, which can be computed by past − now . If the fine-structure constant really is a constant, then any experiment should show that \frac{\ \Delta \alpha\ }{\alpha} ~~ \overset{\underset{\mathsf{~def~}}{}}{=} ~~ \frac{\ \alpha _\mathsf{past} - \alpha _\mathsf{now}\ }{\alpha_\mathsf{now}} ~~=~~ 0 ~, or as close to zero as experiment can measure. Any value far away from zero would indicate that does change over time. So far, most experimental data is consistent with being constant, up to 10 digits of accuracy.
Past rate of change The first experimenters to test whether the fine-structure constant might actually vary examined the
spectral lines of distant astronomical objects and the products of
radioactive decay in the
Oklo natural nuclear fission reactor. Their findings were consistent with no variation in the fine-structure constant between these two vastly separated locations and times. The most recent constraint from Oklo, from Davis & Hamdan (2015), set an upper limit of 11 ppb difference at 95% confidence level, a constraint comparable in strength to that from atomic-clock measurements. Improved technology at the dawn of the 21st century made it possible to probe the value of at much larger distances and to a much greater accuracy. In 1999, a team led by John K. Webb of the
University of New South Wales claimed the first detection of a variation in . Using the
Keck telescopes and a data set of 128
quasars at
redshifts , Webb
et al. found that their spectra were consistent with a slight increase in over the last 10–12 billion years. Specifically, they found that \frac{\ \Delta \alpha\ }{\alpha} ~~ \overset{\underset{\mathsf{~def~}}{}}{=} ~~ \frac{\ \alpha _\mathrm{prev}-\alpha _\mathrm{now}\ }{\alpha_\mathrm{now}} ~~=~~ \left(-5.7\pm 1.0 \right) \times 10^{-6} ~. In other words, they measured the value to be somewhere between and . This is a very small value, but the error bars do not actually include zero. This result either indicates that is not constant or that there is experimental error unaccounted for. In 2004, a smaller study of 23 absorption systems by Chand
et al., using the
Very Large Telescope, found no measurable variation: \frac{\Delta \alpha}{\alpha_\mathrm{em}}\ =\ \left(-0.6\pm 0.6\right) \times 10^{-6} ~. However, in 2007 simple flaws were identified in the analysis method of Chand
et al., discrediting those results. King
et al. have used
Markov chain Monte Carlo methods to investigate the algorithm used by the UNSW group to determine from the quasar spectra, and have found that the algorithm appears to produce correct uncertainties and maximum likelihood estimates for for particular models. This suggests that the statistical uncertainties and best estimate for stated by Webb
et al. and Murphy
et al. are robust. In 2007, Khatri and
Wandelt of the University of Illinois at Urbana-Champaign realized that the
21 cm hyperfine transition in neutral hydrogen of the early universe leaves a unique absorption line imprint in the
cosmic microwave background radiation. They proposed using this effect to measure the value of during the epoch before the formation of the first stars. In principle, this technique provides enough information to measure a variation of 1 part in (4 orders of magnitude better than the current quasar constraints). However, the constraint which can be placed on is strongly dependent upon effective integration time, going as . The European
LOFAR radio telescope would only be able to constrain to about 0.3%. used the frequency ratio of and in single-ion optical atomic clocks to place a very stringent constraint on the present-time temporal variation of , namely = per year. A present day null constraint on the time variation of alpha does not necessarily rule out time variation in the past. Indeed, some theories that predict a variable fine-structure constant also predict that the value of the fine-structure constant should become practically fixed in its value once the universe enters its current
dark energy-dominated epoch.
Spatial variation – Australian dipole Researchers from Australia have said they had identified a variation of the fine-structure constant across the observable universe. These results have not been replicated by other researchers. In September and October 2010, after released research by Webb
et al., physicists
C. Orzel and
S.M. Carroll separately suggested various approaches of how Webb's observations may be wrong. Orzel argues that the study may contain wrong data due to subtle differences in the two telescopes. Carroll takes an altogether different approach: he looks at the fine-structure constant as a scalar field and claims that if the telescopes are correct and the fine-structure constant varies smoothly over the universe, then the scalar field must have a very small mass. However, previous research has shown that the mass is not likely to be extremely small. Both of these scientists' early criticisms point to the fact that different techniques are needed to confirm or contradict the results, a conclusion Webb,
et al., previously stated in their study. == Anthropic explanation ==