Estimates of the association between ambient ozone pollution and cardiorespiratory mortality risk for over 50 US cities, with and without control for seven chemical components of particulate matter. The diagonal line shows where points would fall if effect estimates are exactly the same with and without control for that component of particulate matter. Points fall very close to the diagonal reference lines, indicating there is no evidence that these components confound the relationship between ozone and human mortality. Each circle shows the estimated effect for a single community. Circle size indicates certainty, with larger circles indicating greater certainty. The white diamonds show the pooled estimate across all communities.