They're from the Pitch/FX section of fangraphs.com. If I had to guess I'd say the front of the plate but I don't know what part of the plate they use for this. There's very little difference between the front and back of the plate as far as location even with an extremely hard breaking curve only moving up to 4 inches as it goes from the front to back of the plate. A fastball might sink an inch or two at the most across the plate and would tail far, far less. A pitcher throwing across his body as much as Weaver, and nobody else comes close to him, might get somewhere close to an inch of difference horizontally from front to back. These are quick, back of the envelope calculations. I think I did more detailed calculations somewhere else on here in the past. The curve assumption is 6' of total drop so it would have to hit the plate to get that much drop as it crosses the plate. Anything else is going to be dropping much more slowly and a curve that ends up in the strike zone will be moving a lot less as it crosses the plate. BTW, if one were to take into account the triangular shape of the back part of the plate then the movement in almost every case as the ball crosses the plate would be less.