Abstract
Pedagogical explanation of general covariance is given. I refine the notion of the observer and that of physical quantities to avoid the historical confusion. On the basis of the refinement, I explain the non-trivial usage of the word 'covariance.' I warn confused usage often found in the literature. It is pointed out that experimental predictions, such as of classical tests, aer non-covariant in the nontrivial usage. In this respect, general relativity differs from usual gauge theories, in which every prediction is gauge invariant. I explain the origin of the difference.
