Study design

1. Goats were assigned at random to pens. But complete randomisation can increase the difficulties of day-to-day management of the experiment. In contrast, grouping goats receiving the same feeding regime together in the same block of pens would simplify feeding and reduce possible errors. This might be suitable for an experiment in a well controlled environment provided that there are insignificant variations associated with the position of a pen within the house. Experimental design can thus become a compromise between what is theoretically desirable and what is practically manageable and may well depend on the availability and quality of experimental attendants.

2. It might be possible to design the experiment as a randomised block with genotypes and feeding regimes randomised within blocks in order to accommodate variations caused by the positions of the pens within the building. Pens close to the edge of the house may be more influenced by outside air flows than those inside. This would not help with ease of experiment management but may reduce the size of the residual variance.