Mr. Michael O'Sullivan:

Yes, we did some random sampling and looked at 2,136 cases in connection with recording errors. We looked at all the checkpoints, which amounted to 3,971, and these had over 50 breath tests recorded, which we thought was high. We also looked at 1,984 based on an analytic formulation and a total of 2,134 seemed not to be right, though we could not be sure why. It may have been because of recording issues so we returned them for further examination in the regions.