The method used in Y.-T. Siu's paper on the very-ampleness part of Fujita's conjecture is discussed.