If you're amenable, honourable member...it's not random testing. This is targeted testing, based on analyses that are determined as to higher-risk products and lower-risk products, and those targets are set and reviewed, depending on findings. There is an ongoing flexibility in the system.
For example, you made specific reference to canned testing. Canned testing can be in regard to can integrity; it can be sampling of specific products for specific risks associated, whether they be biological pathogens or other things. The nature of the very programming is such that you have an expected level of find. If it exceeds that level of find, you up the ante, and that testing is then immediately increased.
In a number of our programs—a good example would be fish inspection—we have an ongoing level of detection based on demonstration of the country of its compliance. If we find a shipment out of compliance, we go to 100% testing for the next 10 shipments. That's the very nature of the program. It has to be responsive to what it finds.