Human torso models with an artificial lung and/or a spine resembling to shapes of a standard Japanese male were made up. Resistivities of these artificial components were the same as the ones of canine tissues. The specific resistance of canine spine was considerably higher than lung. The models were filled with 0.2 per cent of sodium chloride. Changes of surface potential on the models were investigated with regard to lead vector using a dipole as an electric source. All these calculations were performed in comparison with a human torso model without the components (homogeneous human torso model). By the experiments, following results were obtained. 1. Human torso model with an artificial lung model (lung model) gave significant changes on image surface whereas the one with a spine model (spine model) produced little change, when a dipole was set at the cardiac center. The tendency became clear on the cardiac center level. Lead vectors for the lung model decreased in lung region whereas they increased in cardiac region on the body surface of the model. 2. When changes in magnitude and direction of lead vector taken with Frank's and McFee-Parungao's systems were studied by setting a dipole in 15 different points within a space where heart was supposed to be present, lesser normality and orthogonality were observed in the lung model than in the spine model. This was clear in the former's system. Magnitudes of lead vector in the lung model decreased in Ⅰ, V(5) and V(6), on the other hand they increased in Ⅱ, Ⅲ, (a)V(F), V(1), V(2) and V(3) of 12 standard leads system, when compared to the ones in homogeneous human torso model. These changes were scarcely observed in the spine model. These results could be summarized as follows: changes of body surface potential produced by electrical hete rogeneity were more remarkably influenced by setting a lung model which has a large volume, than a spine model which has a high resistivity.