My question relates to recording and reading from magnetic tape.  I
have found several references which explain the process in general. 
It seems intuitively obvious that the read gap length should provide a
sin(x)/x filtering of the recorded signal, at least on a first order
approximation.  But what effect does the WRITE gap length have?  This
seems to be a very complicated situation and I've found conflicting
data suggesting what effect the write head gap length has on frequency
response.

What I'm looking for is an approximate transfer function for the
magnetic recording process which depends on (and therefore illustrates
the effects of) both read and write gap length.

A more detailed transfer function would be fine, as long as the effect
of the write gap length is evident, or else a clear explanation of why
the write gap length cannot be simply related to frequency response.