What the above posters have said is correct, and there is some basic physics that can be used to predict a driver's characteristics. If you're not technically inclined, I apologize in advance. This is as simple as I know how to make it while still presenting the basic math involved.
When the wavelength of sound being reproduced is longer (lower in frequency) than the circumference of the moving part of the driver (the piston), the radiation pattern of a driver in free space is approximately spherical. When mounted to a baffle, the radiation pattern is about 180 degrees. As the frequency rises (wavelength gets smaller), the radiation pattern gets narrower. When the wavelength of the sound is about half of the circumference of the driver, the driver starts to "break up", indicated by ripples in the frequency response. At one third of the circumference of the driver, which just happens to be about the diameter of the driver, the response is down about 7dB at 60 degrees off axis. This represents a practical limit on the upper frequency of the driver. There are techniques that can be used to extend this range, but for the sake of discussion assume this is a conventional driver.
As a practical example, take a Scanspeak 18W/8545, a popular 7 inch midwoofer. You can view its data at http://www.d-s-t.com/scs/index.htm. From its datasheet, the pistonic area is 145 square cm. This translates into a diameter of 5.35 inches, and a circumference of 16.8 inches, or 1.4 feet. The speed of sound is approximately 1130 ft/sec, so 1.4 feet corresponds to a wavelength of 1130/1.4=807Hz. If you look at the 8545's frequency response graph you'll see a bump at around 800Hz, and a downward slope from that point upward. At 1600Hz (half the wavelength of 800Hz), the onset of cone breakup is evident. At 2400Hz, the response at 60 degrees off axis is down about 6 dB, which is very close to the predicted value.
Depending on the slope of the crossover, this driver should probably be limited to about a 2KHz maximum crossover frequency. Higher order crossovers can push the limit, while lower order crossovers should be around 1KHz or so. Since it is a rare tweeter that can be crossed over at 1Khz, a higher order crossover will have to be used if you want to make a decent sounding 2-way. Also, while you'd like to have the crossover above 3KHz to get it out of the most audible range, the breakup modes of the driver prevent you from doing that. It would sound too distorted.
If you had a driver with a 4" piston diameter (a roughly 5.25" driver), the 3x wavelength would be about 3200Hz, which would allow the crossover frequency to be moved upward. However, the bass response would suffer, and would likely require a 3-way system instead of a 2-way.
What does all of this have to do with the sound, you ask? If you only use drivers in their linear range, it is pretty much impossible to cover the 20-20KHz range with only 2 drivers. However, you can certainly do 40-20KHz, which is good enough for most people. Full range sound requires at least 3-way operation. But the more crossovers and drivers you have, the more expensive the speaker, and the harder it is to get the drivers to blend well.
If you'd like to read more about this subject, check out "High Performance Loudspeakers" by Martin Colloms.
When the wavelength of sound being reproduced is longer (lower in frequency) than the circumference of the moving part of the driver (the piston), the radiation pattern of a driver in free space is approximately spherical. When mounted to a baffle, the radiation pattern is about 180 degrees. As the frequency rises (wavelength gets smaller), the radiation pattern gets narrower. When the wavelength of the sound is about half of the circumference of the driver, the driver starts to "break up", indicated by ripples in the frequency response. At one third of the circumference of the driver, which just happens to be about the diameter of the driver, the response is down about 7dB at 60 degrees off axis. This represents a practical limit on the upper frequency of the driver. There are techniques that can be used to extend this range, but for the sake of discussion assume this is a conventional driver.
As a practical example, take a Scanspeak 18W/8545, a popular 7 inch midwoofer. You can view its data at http://www.d-s-t.com/scs/index.htm. From its datasheet, the pistonic area is 145 square cm. This translates into a diameter of 5.35 inches, and a circumference of 16.8 inches, or 1.4 feet. The speed of sound is approximately 1130 ft/sec, so 1.4 feet corresponds to a wavelength of 1130/1.4=807Hz. If you look at the 8545's frequency response graph you'll see a bump at around 800Hz, and a downward slope from that point upward. At 1600Hz (half the wavelength of 800Hz), the onset of cone breakup is evident. At 2400Hz, the response at 60 degrees off axis is down about 6 dB, which is very close to the predicted value.
Depending on the slope of the crossover, this driver should probably be limited to about a 2KHz maximum crossover frequency. Higher order crossovers can push the limit, while lower order crossovers should be around 1KHz or so. Since it is a rare tweeter that can be crossed over at 1Khz, a higher order crossover will have to be used if you want to make a decent sounding 2-way. Also, while you'd like to have the crossover above 3KHz to get it out of the most audible range, the breakup modes of the driver prevent you from doing that. It would sound too distorted.
If you had a driver with a 4" piston diameter (a roughly 5.25" driver), the 3x wavelength would be about 3200Hz, which would allow the crossover frequency to be moved upward. However, the bass response would suffer, and would likely require a 3-way system instead of a 2-way.
What does all of this have to do with the sound, you ask? If you only use drivers in their linear range, it is pretty much impossible to cover the 20-20KHz range with only 2 drivers. However, you can certainly do 40-20KHz, which is good enough for most people. Full range sound requires at least 3-way operation. But the more crossovers and drivers you have, the more expensive the speaker, and the harder it is to get the drivers to blend well.
If you'd like to read more about this subject, check out "High Performance Loudspeakers" by Martin Colloms.