Murakami (1987a) developed the hierarchical component analysis maximizing the sum of alpha coefficients of second order composites as a method for analyzing multiset data. The present paper offers some additional information on his method. First, three differnt types of hierarchical component analysis methods are formulated and compared with the present method. The main features of the method, that is, the scale free property and the interpretability of the second order composites are stressed. Second, three data sets, artificial and real, are analyzed by the method. The results showed that the fidelity and the bandwidth of the first order composites can be balanced according to expectation through changing the number of second order composites.