Abstract
Precision cosmology challenges many aspects of fundamental physics. In particular,
quasar absorption lines test the assumed constancy of fundamental constants
over cosmological time-scales and distances. Until recently, the most reliable technique
was the alkali doublet (AD) method where the measured doublet separation
probes variations in the fine-structure constant, αΞ e2/ħc. However, the recently
introduced many-multiplet (MM) method provides several advantages, including a
demonstrated ≈10-fold precision gain. This thesis presents detailed MM analyses
of 3 independent Keck/HIRES samples containing 128 absorption systems with
0.2 > zabs > 3.7. We find 5.6 σ statistical evidence for a smaller α in the absorption
clouds: Δα/α = (-0.574 ± 0.102) x 10-5. All three samples separately yield
consistent, significant Δα/α. The data marginally prefer constant dα/dt rather
than constant Δα/α. The two-point correlation function for α and the angular
distribution of Δα/α give no evidence for spatial variations. We also analyse 21
Keck/HIRES Si iv doublets, obtaining a 3-fold relative precision gain over previous
AD studies: Δα/α = (-0.5 ± 1.3) x 10-5 for 2.0 > zabs > 3.1.
Our statistical evidence for varying α requires careful consideration of systematic
errors. Modelling demonstrates that atmospheric dispersion is potentially important.
However, the quasar spectra suggest a negligible effect on Δα/α. Cosmological
variation in Mg isotopic abundances may affect Δα/α at zabs > 1.8. Galactic
observations and theory suggest diminished 25;26Mg abundances in the low metallicity
quasar absorbers. Removing 25;26Mg isotopes yields more negative Δα/α values.
Overall, known systematic errors can not explain our results.
We also constrain variations in y Ξ α 2gp, comparing H i 21-cm and millimetrewave
molecular absorption in 2 systems. Fitting both the H i and molecular lines
yields the tightest, most reliable current constraints: Δy/y = (-0.20±0.44)x10-5
and (-0.16±0.54)x10-5 at zabs = 0.2467 and 0.6847 respectively. Possible line-ofsight
velocity differences between the H i and molecular absorbing regions dominate
these 1 σ errors. A larger sample of mm/H i comparisons is required to reliably
quantify this uncertainty and provide a potentially crucial check on the MM result.