A test-engineer at Kodak contacted me by email. He read this thread, and Kodak decided to run some tests. They ran samples from several batches of film which they had stored. These batches were dated before, during, and after the failing batches reported here. All turned out fine! He even sent me the graphs. All were fine.
Originally Posted by GraemeMitchell
I wanted to ask him to run tests of film purchased on the open market. Both Michael R and I saw this problem, and Kodak didn't, and that suggests to me that this issue is related to storage-conditions. But the last email sounded like Kodak had decided that everything was fine, and wasn't going to investigate this more, and that's how the story ended.