The flux profile of the Perseid stream from 1988 to 1994 is analyzed using visual observations of meteors. The position and peak flux of recently detected outburst maxima are found for each year, as are the location and magnitude of the regular maximum. Details of the reduction procedure for these visual data are also presented. The Perseid stream is found to consist of three primary components: a long-lived and relatively weak background component, a core component which is active for 1-2 days near the main peak of the shower, and an outburst component which is active for only a few hours. The latter component changes position and strength dramatically from year to year, while the core maximum remains at a near constant location of lambda(z) = 139.degrees 96 +/- 0.degrees 05 and shows a variance of less than 30% from a mean peak flux of S-6.5 max = 96 +/- 16 meteoroids per 10(9) km(3) during 1988-1994. The outburst peak locations show no systematic variation from year to year, though the magnitude of the peak flux shows a significant difference for returns before 1991 compared to those after 1991. This may indicate an ejection origin other than 1862 for outburst material encountered before 1991. The particle composition of the outburst maximum and that of the core maximum are indistinguishable. A double maximum in the population index associated with these activity maxima has been detected in the average Perseid profile. High temporal resolution study of the flux profile of the ascending branches of the outburst profiles in 1993 and 1994 suggests that this component may consist of sub-components of differing ages. The mean visual flux of the shower is in excellent agreement with past radar results. The most important features of the stream gleaned from this study which must be explained by any model of the Perseids are also summarized. (C) 1996 Academic Press, Inc.
