I would suspect that the canceling machine had an envelope "jam" near the cancelling head, and the envelope following this one was struck a number of times before the jam was cleared. After continuing down the transport it came in contact with this envelope while the excess ink was still wet..
When a canceling machine is working properly, each letter passing through it trips the inked canceling head, which then begins to rotate and make an imprint on the stamp, then stop in a neutral position and wait for the next letter to come through.
When the machine is out of adjustment, like having a spring broken or a similar defect in the tripping mechanism, the canceling head portion of the machine may continue to rotate and print repeatedly onto the rubber roller which rolls along the back side of the envelope. The backside roller picks up ink and then deposits it on the back of envelopes passing through later.
Add: Notice the lettering, dial, and wavy lines imprint of the backside offset are ghosted, i.e., they are white, and just inked around the edges, where the inked canceling portion of the metal die pushed into the soft rubber roller and squeezed the ink to the edges.
Yes, that is my belief. Although I would use the terms "front roller" (the canceling head) and "back roller" as the letter is vertical and is sliding along on its top side, up side down, as it passes through the machine.
More specifically, to accommodate envelopes of differing heights and get the postmark the same distance down from the top (as we see it), everything is actually done upside down. The stream of faced envelopes is fed up side down into the machine and the cancel is also applied up side down. The result is that "two up side downs make a right side up" with all cancels the same distance down from the top of each letter.