Spikes versus Rubber on wood floor?

If the floor seems to be the offending item, I always recommend fixing the problem, instead of adding some other item into the mix to try to mitigate the floor problem.

Nsgarch has some good advice about locating the speakers over the joists, and there can be supports placed under the joists too.

I had speakers with cones and down firing subs sitting on a suspended floor in a relatively new home. The floor has a resonant frequency of 34 hz, and I know because at that frequency the amplitude of the vibrations increases so much it shakes the chair and tickles my butt!

To solve it, I cross braced the joists, put a post under the center of the floor, and put a piece of slate between the cones and the carpet (isolation). This combination works very well indeed, and there is no visible movement of the speaker tops.

So, my experience and that of Fiddler are very similar, in that we both found isolation was necessary on a suspended floor. That does not mean coupling should never be used, as these same speakers on a concrete floor shine with just the cones.

Obviously even well constructed speakers vibrate or spikes would not work their way right through pennies placed underneath them.
Given all the divergence in answers, I decided to approach this logically through a hypothetical example. Please tell me where you think my reasoning is correct or incorrect. Lets say we place a vibrator on a kitchen table when the kids aren't home. In my example the vibrator is equivant to a speaker.
We turn it on (turn about is fair play). What will it do? It will clatter. This is equivilant of a poorly coupled speaker. Then we press down on it adding weight. Now it is coupled to the table and the table will act as an amplifier of the vibration - rather the disipate, the noise will get louder, true? This would be similar to a coupled speaker interacting with the wood floor. The speaker is now less vibratory but the table is magnifiying the sound.
Next we place an inert substance such as styofoam, sand, or rock between the vibrator and the table. We now get less noise. This suggests to me the best is to firmly couple the speaker to something rigid enough to reduce the speaker's vibration,and under this have something non-resonant so that the vibration is not passed to the floor.

Zargon, your case is the special exception I mentioned earlier, namely: you have a transducer (subwoofer) that moves up and down, same as the (only) direction the floor can move. So naturally it can mechanically excite the floor (grab it and shake it) in that direction at 34 Hz or whatever. But a normal forward facing transducer can't do that because the floor can't respond to its forward-backward motion. A woofer or sub can of course possibly excite strong standing waves in the air in a room (not the speakers' fault, just a factor of the room dimensions) and these standing waves can excite the floor to resonate. My point is that to decide how best to accomodate a speaker, one has to look at how it might want to move and then figure out how best to restrain it.

Gammajo's analogy doesn't really apply to loudspeakers for two reasons. First a vibrator's action is usually rotational (semi-omnidirectional), and loudspeaker transducers are reciprocal, whether they're cones or electrostats. Second, a vibrator is designed to transfer energy to another solid (your body). But loudspeaker transducers must absolutely NOT do that (transfer any energy to the enclosure.) ALL their motion must go into vibrating the air. That is why the speaker enclosure or frame must not vibrate (resonate) or be able to physically move (like rock back and forth on rubber pads or a rug or a too-flexible floor.)

.