Observations of the diurnal variations of OClO and BrO during austral spring, 1987 using long-path visible and near-ultraviolet absorption spectroscopy are presented and compared to simplified model calculations. It is shown that care must be taken to compare model calculations and measurements along the line of sight of the instrument. Evening twilight observations of OClO are found to be broadly consistent with current photochemical schemes, assuming ClO and BrO levels near 50 mbar of about 0.5 parts per billion by volume (ppbv) and 7 parts per trillion by volume (pptv), respectively, throughout the observing period from late August to mid-October. Nighttime observations of OClO obtained using the Moon as a light source display evidence for growth after sunset in late August, but not in late September. Further, the observed morning twilight OClO abundances are in agreement with model calculations in late August, but they generally fall below calculations in late September and October. Observations in BrO in mid-September show far greater evening than morning twilight abundances. It is shown that the diurnal variations of BrO and OClO in mid-September and October can be explained by formation of the BrONO2 reservoir species at night, although other reservoir species with comparably long lifetimes could also explain the observations. If formation of BrONO2 is the correct explanation for these data, the observations suggest that NO2 levels in the Antarctic lower stratosphere are of the order of a few pptv or less in late August, a few tens of pptv in mid- September, and a few hundred pptv in October. ¿ American Geophysical Union 1989