The role of the seam in the ‘swing’ of a cricket ball is investigated via unsteady force and surface-pressure measurements and oil-flow visualization in a low-turbulence wind tunnel. Various seam angles of the ball and flow speeds are considered. Static tests are carried out on a new ‘SG Test’ cricket ball as well as its idealized models: a smooth sphere with one and five trips. To study the effect of surface roughness of the ball as the game progresses, force measurements are also carried out on a cricket ball that is manually roughened, on one-half and completely, to model a ball that has been in play for approximately 40 overs (240 deliveries/balls). The Reynolds number ($Re$) is based on the free-stream speed and diameter of the respective model. A new cricket ball experiences three flow states with increase in $Re$: no swing (NS), conventional swing (CS) and reverse swing (RS). At relatively low $Re$, in the NS regime, the seam does not have any significant effect on the flow. The separation of the laminar boundary layer, with no subsequent reattachment, is almost axisymmetric with respect to the free-stream flow. Therefore, the ball does not experience any significant lateral force. Beyond a certain $Re$, the boundary layer on the seam side of the ball undergoes transition. The boundary layer on the non-seam side, however, continues to undergo a laminar separation with no reattachment, thereby creating a lateral force in the direction of the seam, leading to CS. The onset of the CS regime is marked by intermittent formation of a laminar separation bubble (LSB) on the surface of the ball in the region between the laminar separation of the boundary layer and its reattachment at a downstream location. Owing to the varying azimuthal location of the seam, with respect to the front stagnation point on the ball, the transition via LSB formation is localized to a specific region over the seam side. In other regions, the boundary layer either transitions directly without the formation of an LSB, or separates on encountering the seam with no further reattachment. The spatial extent of the region where the flow directly transitions to a turbulent state increases with increase in $Re$, while that of the LSB decreases. Interestingly, the flow dynamics is such that the magnitude of the swing force coefficient stays relatively constant with increase in $Re$. With further increase in $Re$, the boundary layer on the non-seam side undergoes a transition via formation of an LSB. This, along with an upstream shift of the separation point on the seam side, leads to a switch in the direction of the lateral force. It now acts away from the seam, and leads to RS. The transition from CS to RS occurs over a very narrow range of $Re$ wherein the flow intermittently switches between the two flow states. It is observed that the transition of the boundary layer on the seam side leads to an upstream shift of the separation point on the non-seam side at the onset of CS. A complementary effect is observed at the onset of RS. Experiments on a ball that is manually roughened bring out the relative effect of the seam and roughness on the transition of the boundary layer. Compared to a new ball, the magnitude of the maximum swing force coefficient for a rough ball is smaller during the CS regime, and larger during the RS regime. Unlike other models, the ball with roughened non-seam side and smooth seam side, for certain seam orientations, exhibits RS at relatively lower speeds and CS at higher speeds. The forces measured on the cricket ball are utilized to estimate the trajectory of the ball bowled at various initial speeds and seam angles. The lateral movement of the ball depends very significantly on the seam angle, surface roughness and speed of the ball at its delivery. The maximum lateral deviation of a new ball during RS is found to be less than half of that observed in CS. On the other hand, the lateral movement of a roughened ball during RS may significantly exceed its movement during CS. The range of the speed of the ball, for various seam orientations and surface roughnesses, are estimated wherein it undergoes CS, RS or one followed by the other. Optimal conditions are estimated for the desired lateral movement of the ball.