Well, when I did it, leak testing was a 2-phase process.
Let's say the requirement was a leak rate of < 1 x 10^-5 std cc/sec. for the entire joint.
First, we would skim around the joint we were concerned about to detect whether or not there was a leak or leaks (detection).
Second, if we detected a leak(s) at a certain spot on the joint, we would pause there to read the leak rate (measurement).
If we found more than 1 leak we added them together and compared it to the spec.
IIRC, the leak tester could read down to 3 x 10^-7.
To validate the tester, we would skim by the standard leak (whose leak rate was smaller than the max allowed leak of the UUT) and verify we could detect it that way, then hold it steady to measure the leak rate and compare our measurement to the standard’s leak rate