ABSTRACT
The Wishart distribution was first derived by Fisher (1915) for p=2. Wishart (1928) gave a geometrical derivation of this distribution for general p. Ingham (1933) derived this distribution from its characteristic function. Elfving (1947), Mauldan (1955), and Olkin and Roy (1954) used matrix transformations to derive the Bartlett decomposition of the Wishart matrix from sample observations and then derived the distribution of the Wishart matrix. Khirsagar (1959) used random orthogonal transformations to derive the Wishart distribution and the distribution of Bartlett decomposition. Sverdrup (1947) derived this distribution by straightforward integration over the sample space. Narain (1948) and Ogawa (1953) used the regression approach, Ogawa’s approach being more elegant. Rasch (1948) and Khatri (1963) also gave alternative derivations of this distribution.
