The flame expansion process (``light-round'') during the ignition transient in annular combustors depends on a number of parameters such as equivalence ratio (and hence laminar burning velocity, $SL$, of the mixture), turbulent intensity, mean flow magnitude and direction, geometry, and spark location. Here, an experimental study on a fully premixed, swirled, bluff-body stabilised annular combustor is carried out to identify the sensitivity of the light-round to these parameters. A wide range of conditions were assessed: two inter-burner spacing distances, two fuels (methane and ethylene), bulk velocities from 10 to 30 m/s, and $\varphi$ between 0.75 and 1 for methane and 0.58 and 0.9 for ethylene. The spark location was varied longitudinally ($x/D$ = 0.5 and $x/D$ = 5, where $D$ is the bluff body diameter, expected to lie inside and downstream of the inner recirculation zone of a single burner, respectively) and azimuthally. The propagation of the flame during the ignition transient was investigated via high speed (10 kHz) OH$\ast$ chemiluminescence using two cameras to simultaneously image the annular chamber from axially downstream and from the side of the combustor. The pattern of flame propagation depended on the initial longitudinal spark location and comprised of burner-to-burner propagation close to the bluff bodies and upstream propagation of the flame front. The spark azimuthal position\textcolor{red}{, in this horizontal configuration,} had a negligible impact on the light-round time ($\tau LR$), thus buoyancy plays a minor role in the process. In contrast, sparking at $x/D$ = 5 resulted in an increase in $\tau LR$ by $\sim$30-40% for all the conditions examined. The inter-burner spacing had a negligible effect on $\tau LR$. When increasing bulk velocity, $\tau LR$ decreased. For a constant bulk velocity, $\tau LR$ depended strongly on $SL$ and it was found that mixtures with the same $SL$ from different fuels resulted in the same $\tau LR$. Further, the observed propagation speed, corrected for dilatation, was approximately proportional to $SL$ and was within 30% of estimates of the turbulent flame speed at the same conditions. These findings suggest that $SL$ is one of the controlling parameters of the light-round process; hence turbulent flame propagation has a major role in the light-round process, in addition to dilatation and flame advection by the mean flow. The results reported in the study help explain the mechanism of light-round and can assist the development of efficient ignition procedures in aviation gas turbines.