Who doesn't enjoy a conversation on this topic? Critique of any error invited:
taurussable at comment #1: Sampras had to transfer some weight onto his back foot in order to turn back. That was the point of the 'chong,' taught to him and others by Del Little at the Kramer Club and later at Racquet Swingers in Lomita, CA. The chong is the platform stance with the back foot at a steep angle to the front foot, pointing back, not too far apart, heels pointed at each other approximately, with an angle between the feet of 70 to as much as 90 degrees.
Generally: Pete used a lot of hip shift, dip, which occured after his turn back. It is impossible to keep the weight back during that shift, so though his back foot was on the ground, his weight had to have shifted a good bit and therefore his UB was already 'wanting' to rotate forward. That's what happens when the UB is suddently bearing on the leg with which it is not aligned. No?
SA at #6: You can actually see RF's back foot push-up propel his hips, UB, around, start them moving. Pete's hips can be seen moving off his front leg launch. But it is also true that Pete starts his rotation with the leg launch and tossing arm drop (backwards at first until rotation makes it 'to the side') delaying his racquet drop just a bit, but not in the extreme a la Roddick.
Nostradamus at #15: Pete has already approached complete hip shift before pushing left with his right foot as a final bit of front leg loading, no? There is, at that point, little weight to shift, but a leg launch to 'trigger' with that little push which comes just with Pete's "hip dip," that characteristic move at the end of his hip shift, at which point he launches. Pete's always been (until some recent exhibitions...) a toe dragger. He actually was keeping his feet closer together (relatively) at a local exhibition here four years ago, Rosemont College, but still angled.
Chas Tennis at #41: That's interesting, the 'why' of those who push up and a bit forward off the right foot in the launch (I think of Federer) before their left foot full launch: If the right hip is lifted and pushed a bit forward, the UB is no longer inhibited by it from rotating, and the high hip encourages the free rise of the hitting shoulder and drop of the tossing shoulder, agree. The hitting shoulder rise is vital because it connects the arm to the UB: The UB's momentum is transfered via the shoulder to the hitting upper arm, and that powers the hitting elbow up before the arm has much of a say in the matter. The shoulder rise powers the hitting arm up at a critical instant, boosting the effect of the UB rotation and the racquets moment of free swing. No?
SystemicAnomaly at comment #43: "However, the act of pulling the arm down and tucking it into the body will help by reducing his rotational inertia so that his cartwheel action is faster." Perhaps you meant "the act of throwing/letting fall the tossing arm creates in it inertia/momentum, so that when he pulls it in (shortens the lever arm) the UB will receive the inertia, and just like a figure skater, the UB will rotate faster." ---Absolutely agree. The drop and pull-in is an essential ingredient of a pro serve. It is not just the momentum transfered to the UB, but the ease of controlling the timing of that momentum transfer to coordinate the UB timing to the key moment of hitting elbow rise. The pull in only provides acceleration to the UB if the arm (and UB) are effectively rotating as well as falling, but the rotation element can be small, yes?
SystemicAnomaly at comment #45: Yep. If you have some higher frame-rate Roddick serves you'll see that he doesn't just drop the tossing arm. He pulls it a bit toward the back fence just before launch and literally throws it toward the net and down. (Of course you'll also see that his feet are off the ground before he has even completed the toss back/down of his racquet on his first serve. I can still remember the day I first noticed that. I was shocked.)
General: Jimmy Connors and the platform serve:
Connors was 10 years old when the foot fault rule changed in 1961, and was in the first group to develop a serving motion strategy that optimized the permission to be off the ground before contact, and to land in the court after contact. So it's natural he was an early platform server: the history of the rule change:
"It used to be that a tennis player while serving the ball could be moving––walking or running toward the service line. Furthermore he or she could use this running start to advance quickly to the net as part of the service motion. That all changed beginning in 1903.
The newly proposed rule stated “It is not a fault if one only of the server’s feet do not touch the ground at the moment at which the service is delivered. He shall place both feet on the ground immediately before serving, and not take a running or a walking start.” (New York Times, June 7, 1903).
This rule was arbitrated by the English who did not care for this practice employed by American players who often appeared to arrive at the net before the ball did. The Brits got this rule in place to quell the U.S. enthusiasm for rushing the net. So, from 1908 until 1960, a tennis player had to keep one foot on the ground at all times during the service motion. If both feet left the ground it was ruled a foot fault.
The rule was amended in 1961."