The function can be used in several ways:

• butter(n,Wn) - Design an n pole lowpass or bandpass Butterworth filter with corner frequency Wn (scalar or 2 element vector).
• buffer(n,Wn,type) - Design an n pole lowpass, highpass, bandpass, or bandstop filter, with corner frequencies specified by Wn (scalar or 2 element vector).
• butter(...,'s') - If 's' is added to the end of the function, an analog filter will be designed.

The parameter Wn is the relative corner frequency of the filter for digital filters. For analog filters, it is specified in radians per second. The size of the Wn depends on the type of the filter:

• lowpass, highpass - A scalar value representing the corner frequency of the filter.
• bandpass, bandstop - A 2-element vector representing the corner frequencies: [Wn1 Wn2]. Wn1 < Wn2.

The relative frequency in case of digital filters is defined as $ 2\frac{f}{f_s} $, where f is the frequency of interest. Thus 0 means 0Hz, 1 means the Nyquist rate (fs/2).

The parameter type is a string value describing the filter type:

• "low": lowpass (you can omit type)
• "high": highpass
• "bandpass": bandpass (you can omit type)
• "stop": bandstop

For more information, see http://en.wikipedia.org/wiki/Butterworth_filter.