We excavated trenches across a prominent lineament and surface scarp, exposing a major fault with clear evidence of recurrent activity. The main strand of the fault juxtaposes mid-Holocene lake deposits against late Holocene, inter-bedded lake and alluvial deposits. We identified past surface ruptures by the presence of filled fissures, upward fault terminations, folding, angular unconformities, from which we identify evidence for up to eight surface ruptures that have occurred in the past 1700 years. The earliest two events have weak age control, but the past six events, constrained by a combination of C14 and OSL, suggest an average recurrence interval (RI) of 117 ± 55 years and a lapse time of over 210 years. The RI is likely closer to 148 years, however due to event 2 being constrained entirely by C14, it is clustered closer to event 3 on account of inheritance, thereby reducing the RI. The two most recent events are likely the November 22, 1800 and ca 1550 earthquakes, and correlate to events 2 and 3 at Hog Lake, ~50 km to the NW, based on geomorphic offset mapping along the Clark fault. Three earlier events (3, 4 and 5 at Clark Lake) are constrained to have ruptured between about 1359 and 1164 AD and likely correlate to events 4, 6 and 7 at Hog Lake. Event 5 at Hog Lake does not appear to correspond to any events at Clark Lake, suggesting that some of the ruptures in the Hog Lake cluster may correspond to rupture of the northern part of the zone, similar to what occurred in the 1918 M6.9 earthquake. These observations along with geomorphic offset observations suggest that the entire Clark fault, and possibly the Casa Loma fault, fail together in some large earthquakes (Mw7.3) whereas the northern Clark fault may fail more frequently in M6.5-6.9 earthquakes, as occurred in 1899 and 1918.