I figured that to really put these stamps to the test, I should send the letters to an address relatively far away — to make sure it went through a lot of depots, verification centers, biometric drug-sniffers, or whatever. I don’t know how this stuff works; I assumed the barcode encoded a lot of crucial information about where the letter came from, where it was going, and how long the stamp should be honored. So I arranged with friends a thousand miles away (in Seattle) to receive the letters, and as a control subject, sent one letter that night of April 15. The next letter was sent the next day. And so on, at increasing intervals of time, through April 29, a full two weeks after the date of the stamp. I expected that letters sent in the first week or so would arrive, and then they’d start coming back. I was wrong. They all made it.
The interesting part was that, as predicted, not all of the stamps arrived with cancellations. Of the ten sent to Seattle, only six arrived there canceled — meaning that four envelopes (40%) arrived indicating only the April 15 date and no other postmark.
In the course of his experiment he discovered that the letters that arrived uncancelled could be remailed, and were accepted a second time by post office processing equipment. Doing that is strictly against postal regulations; you are not allowed to reuse "skips" (stamps not cancelled in transit). But his observation that an item can be posted after the date printed on the stamp is potentially useful in a variety of situations.
Link. Image credit [yes, we know it's an old one...]