Data are presented which enhance the idea that the influenza virus is an unstable agent. They indicate that the O-D type of variation is not a discontinuous mutation but rather is a reversible phenomenon. The O and the D forms of virus both appear to be inherent in the virus particle; the dominance of one or the other form seems to be subject to chance occurrences, but is influenced by the conditions under which the virus is propagated.
The capacity of the O form of virus to agglutinate guinea pig but not chicken crythrocytes is a relative, not an absolute phenomenon; allantoic fluids which exhibit clear-cut O form hemagglutination may be made to exhibit D form merely by addition of suitable buffer to the test system. That point is of importance from the viewpoint of the mechanism of influenza virus hemagglutination.