C2-C5 hydrocarbons were measured in situ at Alert, Nunavut, Canada from April 14 to May 10, 1998 at a minimum frequency of 12 samples per day. During the sampling period, an event occurred where the O3 mixing ratio fell from about 15 ppbv to the instrumental detection limit of less than 1 ppbv in a timespan of about 24 hours. Mixing ratios of alkanes and ethyne decreased concurrently with the O3 decrease. Ancillary data suggested that this was essentially due to chemical reactions only, most probably involving halogen atoms. The hydrocarbon concentrations from the onset of O3 decrease until its virtually complete depletion are used in a first order kinetics analysis to calculate the concentrations of Cl and Br atoms that are required to explain these observations. The Cl atom concentration is calculated to be 7.5¿104 molec.cm-3. Such a concentration is near the upper limit of earlier estimates, but cannot explain the observed decay of ethyne. Assuming the deficit to be due to bromine atom chemistry, the estimated Br atom concentration is 1.4¿107 molec.cm-3. The ethyne mixing ratio continued to decrease after virtual total O3 depletion suggesting that Br atoms persisted when O3 was depleted. Hence it is postulated that the depletion of total gaseous mercury (TGM) which terminates upon virtual complete O3 depletion is due to a reaction between gaseous mercury and the BrO molecule. ¿ 2000 American Geophysical Union