Annett (1970a) described 283 subjects classified for hand preference and measured for the movement time of each hand on a peg moving task. A new sample of 804 subjects has been examined and classified as before. The mean differences between hands for each preference group are very similar in both samples. Hand preference can be regarded as anchored to an objective measure of manual skill. Both preference and skill are continuously distributed variables, not discrete ones. The quest for the 'essence' of left handedness should be replaced by studies of the characteristics of the distributions involved and of the criteria used to identify segments of the distributions.