In some experiments, where the aim is to compare a set of treatments, there are one or two sources of variation that can be accounted for at the design stage of a study. The statistical technique that is used in these situation is blocking and it can be used to reduce the variance of pairwise treatment comparisons.
When designing an experiment with a single blocking factor, a randomised block design (RBD) can be used if there are sufficient resources to investigated all treatments within each of the blocks of the starting design. When this is not possible an incomplete block design is required and there are various incomplete block designs that are available for specific combinations of design parameters.
The general block design investigates a set of v treatments allocated to n experimental units across b blocks. Each of the blocks contains k units and it is not the case that the size of each block has to be the same. However this is a desirable feature as it contributes towards the balance of the design to ensure that all treatment comparisons are made with the same or similar precision.