This standard demonstration which shows that sound is not transmitted by a vacuum has been improved by Prof. Rudnick. A sound transmitting seal over the vacuum jar has been provided so that the bell can actually be heard before the jar is evacuated. Circuitry built by R. Keolian rings the bell in bursts and simultaneously flashes a light so that the students can see that the bell is still ringing after the jar is evacuated and they can no longer hear it (and also note that light is transmitted by a vacuum, even though sound is not!).
Some have pointed out that this demo is not really what it seems. Sound travels very well in a poor vacuum, as long as the mean free path for collisions of air molecules is much less than the size of the container. For pressures achievable with the roughing pumps we use, about 1/10,000 of atmospheric pressure, the mean free path is about 1 mm. The speed of sound is proportional to the square root of the pressure divided by the density. Since the density is proportional to the pressure, the velocity of sound is independent of pressure.
So why does the sound diminish as the air is pumped out? It is a problem of the impedance match between the air and the bell and the air and the glass. The impedance of the air is proportional to the square root of the product of the pressure and density. Pumping out the air reduces the impedance of the air by a factor of 10,000 and the vibrations of the bell are not coupled to the air. If there was sound in the remaining air, it would not couple to the glass but instead be reflected at the interface.