Confusing on-line doc on replace-match

[Jonas Jarnestrom]

In GNU Emacs 20.6.1 (sparc-sun-solaris2.6, X toolkit)

As an addendum to the bug report I sent yesterday in this matter,
I'd like to file a complaint on the on-line description of the
function replace-match.

I find the text confusing regarding specification of replacement
of subexps.
According to text, subexp replacement can be done either with arg5
or with arg1 using '\N' notation. 

I tried to use the arg1 method as alternative to the arg5 dito, 
with little success.
I was unable to get the same behaviour with '\1' in arg1, as with 1
in arg5. Does the arg1 method work? If so, it would be nice with an
illustrative example in the on-line help.

Further, I am unable to grasp how arg4 is supposed to be used,
from the on-line doc. Here it would also have been nice with an
illustrative example.

Regards,
Jonas Jarnestrom, Stockholm, Sweden