Logarithmic Frequency response graphs

Why are frequency response graphs always charted on a logaritmic scale?
Because it tends to make high frequency irregularities look less significant - good marketing.
Because in order to show the same results on a linear scale, the page would have to be 10 feet wide.
Because there are twice as many frequency/cycles for every octive and the graph repeats every decade so that it is readable. I believe this is a semi-logarithmic graph as the vertical scale is linear.
Human hearing is also perceived in a logarithmic scale.
Thanks for the responses. Let me re-phrase the question a little more specifically.

Why do I usually see freq. response graphed logaritmically between octaves while the ocataves themselves are spaces linearly?
Duddley...A Logarithmic scale shows DECADES equally spaced...100, 1000, 10,000 etc. Octaves would be 20, 40, 80, 160, etc.

A Log scale is used when the quantity being plotted behaves in a Logarithmic way. If you plot such a quantity on a linear scale and you pick the scale so that it is appropriate for one end of the range, the other end will be all scrunched up or stretched out, so that it cant't be read. (Take one of your Log scale plots and try to plot it on a linear scale).

If one axis is linear, and the other is logarithmic, the plot is called "Log/Linear".
Thanks, Eldartford, but I must still be missing something because it seems to me that the x axis would not be any harder to read if the octaves between the decades were spaced evenly on the line. Since the octives themsleves are spaced evenly, what part gets scrunched this way?

I appreciate your patience - Ijust know I'm going to wake up in the middle of the night shouting Aha! soon.
Rec got it right. Human hearing and human music is not linear, but logarithmic in nature. Think about it. As an example, there are eight notes and twenty integer frequencies in the 20Hz to 40Hz octave. There are still only eight notes, but 10,000 integer frequencies in the octave from 10,000Hz to 20,000Hz. A good frequency response graph will highlight the musically relevant note info and lessen the visual impact of the less important integer frequency data.
Duddley, what you propose is a piano keyboard-like scaling, as '61 explained. Yes, both it and a traditional decade-log scale are both easy to read, and appropriate for audio.