The booster is securely clamped to the launch stand. However the starship is only resting on top of the booster - the two parts of the ship are not clamped together.

Launch clamps are released on command of the engine computer - i.e when it detects that sufficient engines are running correctly it commands a clamp release.

Note that Starship Booster's outer engines are actually started using feeds from the launch pad. They need power and pressurized nitrogen from ground equipment to start. The outer engines are not able to restart once off the ground.

So the clamps are required to keep the vehicle in place until all the engines that will start have had a chance to do so. They're also useful if there is an abort during engine start so the vehicle remains secure.

> However the starship is only resting on top of the booster - the two parts of the ship are not clamped together.

That isn't what it looked like to me when it made what, 3 go arounds?

I may be incorrect and they are now using explosive bolts. I've seen those mentioned elsewhere.

However the engines were on the whole time, producing >1G of thrust along the direction of the booster. So that force is going to hold the StarShip and the booster together. But you would expect some lateral force from air friction to push the stack apart at some point.

Might just be the thrust-to-weight being under 1 while they light the engines, but it visibly sits there for a signifcant amount of time, much more than you'd get with a landing suicide burn.

Then what holds the rocket down while the engines are spooling up?

Could possibly this explain why there was so much lateral movement of the booster before it cleared the tower?