Summary.Non‐equivalence in the performance of the samples of children used in standardising the several members of a series of similar tests results in lack of comparability between the standardised scores obtained with different tests in the series; all the standardised scores from one test may be positively or negatively biassed relative to those from another. The bias is named ‘zero error’.The zero errors of twenty Moray House verbal reasoning tests, all relative to one taken as standard, are estimated and shown to be larger for more recent than for earlier tests. The increase is ascribed mainly to increased test sophistication in the later standardisation samples.Local education authorities should take zero errors into account in their allocation proce