From RKG on The Firing Line Forum:
The notion that the hammer block is necessary to prevent firing if a cocked hammer is somehow dropped without a finger on the trigger can be readily demonstrated to be false.
Take a S&W revolver and remove the hammer block. Reassemble the revolver. With it completely unloaded, cock the revolver, point it toward the ceiling, and drop a pencil with a fairly new eraser down the barrel, eraser end down. Pull the trigger, and the pencil will jump (sometimes all the way out the barrel), signifying that the pin hit the eraser just as it would have hit a primer.
Now cock it again and drop the pencil again. This time, start tapping on the trigger with something (I use a teaspoon). After a couple of taps, the hammer drops, but the pencil doesn't move.
Still skeptical? OK, cock the revover with the cylinder open (if you know how), put your pinky over the firing pin hole in the recoil shield, and pull the trigger. After putting a bandaid on your bleeding finger, do it again (use the other pinky) this time tapping the trigger with the spoon. Hammer will fall and your second pinky won't feel a thing.
The pencil-and-spoon test simulates a push-off, a jar-off, a sear failure, or even a twig brushing the trigger of some moron's gun as he walks through the wood with it cocked. What happens is that as the trigger falls, the rebound slide moves forward, and it will always get home before the hammer does.
The Navy event was borne of the notion that sufficient force could be applied to the hammer of an uncocked revolver to cause it to fire. This is theoretically possible if (a) the rebound slide fails, (b) the internal hammer spur (the part that rides on the rebound slide) fails, or (c) the hammer pivot pin fails. I seriously doubt that it has ever happened in real life. I happened to witness experiments at Smith & Wesson, and after a lot of pounding on the hammer, all that ever broke was the external hammer spur (the thing you put your thumb on to cock the hammer). Not once could a revolver be made to fire this way.
To take it one step further. Examine the post-Navy hammer block. With the hammer cocked, the hammer block is retracted (i.e., out of the way of the falling hammer). If somehow you hypothesized that the hammer could pushed or jarred off and the trigger did not move forward (e.g., rebound slide spring missing; rebound slide frozen in place), so that the rebound slide did not move forward, the hammer block would remain retracted, since its action is dependent on the movement of the rebound slide (off of which the hammer block cams). If this were to happen, the revolver would fire, notwithstanding the hammer block. (By the way, I am unaware of my hypothetical scenario ever happening in real life.)
I cannot prove this, and no one at Smith & Wesson has ever admitted it (and the folks there at the time are all dead), but I am convinced that Smith added the hammer block simply to placate the Navy and prevent cancellation of a contract.