This is getting weird.
Sorry about the long read, but here goes...
I downloaded the pics in the first post in order to make an animated gif of one card 'turning into' the other one, and something doesn't make sense at all! Granted, the original pics weren't the same size, so I had to resize one to match the other. (It's important to note that any size adjustments were made both laterally and horizontally at the same time. In other words, I didn't simply stretch it side to side to make it fit. It was proportionate.) These are seemingly flat scans (no perspective manipulation), so this method is proper. My areas of concentration were the top red border and the large crease streaking down the right side. I matched up/aligned those two elements as best I could, because the assumption seemed to be that those things were pretty consistent across the pair of pics.
What you see here is the original card laid atop the 'doctored' card with an opacity of 47%. And here is where it gets batsh_t crazy. You see how "Cracker Jack Ball Players," the top red border and the crease itself look virtually clear as day (in other words, not a lot of distortion from one pic being laid onto the other)? Now look at the "Johnson" line. See how the one hangs significantly lower? That card is the original card with all of the creases.
Think about that for a second. If a card is full of creases and you soak it and try to flatten it out to make said creases disappear, the card would get a bit longer, NOT shorter. Here, the EXACT OPPOSITE happened. The card that's significantly 'shorter' is the one with the creases gone. Again, the pics weren't exactly the same size to begin with, etc., etc. (so that could/would definitely account for some of it), but what in heck is going on????
1914cjjohnson2.jpg
Two other things of note. If you look at the width of the card(s) in the scan, they are pretty identical. Since they 'match,' how come the length is so far off?? Again, it makes no sense.
And, separately, I laid the PSA labels on top of each other and they (the words, numbers and bar code) match perfectly. If PSA printed two separate labels with the same info, would they be exactly the same? I would assume yes, if it was within a certain span of time when their methods were unchanged.