ABSTRACT The emission profile of H a above the continuum in the spectrum of fi Lyr appears to consist of two components, a 'broad" one of total width about 1000 km sec-1 and central intensity approximately equal to that of the neighboring continuum at midprimary eclipse, and a "narrow" one of total width between 600 km sec-1 and 700 km sec-1 and central intensity several times that of the broad component. The total intensity of the emission remains approximately constant over time intervals comparable to the orbital period, the large apparent changes being maiuly the results of the changing intensity of the stellar continua during the eclipses. From one year to another, however, real changes in the mean emission intensity may occur. The center of the narrow component shows litfie or no velocity shift throughout the orbital cycle. The center of the broad component shows shifts that could be interpreted as being opposite in phase to those shown by the primary (B8 absorption) spectrum. Preliminary measures of Hfi and H7 suggest that the Balmer decrement of the narrow component is appreciably steeper than that of the broad component. A tentative explanation of these facts is offered. The narrow component of the emission is believed to arise in a shell or ring surrounding the whole system, while the broad component comes from an envelope around the still unobserved secondary star. The "satellite" lines seen in the spectrum during primary eclipse are explained as absorption lines produced in the envelope, rather than as a direct result of absorption by gaseous streams. Actually there is only one gaseous stream in the system flowing from the B8 component toward its companion. Key words: binary star - emission lines - spectrophotometry