Altogether, it is much like (for a different type declaration of p):

		for (p = head; p && p->key < key; p = p->next)

	except that the p in my version is always a pointer to the p in
	this latter version.  Specifically, it is a pointer to the variable
	containing the 'next' pointer in question.  So it can be modified
	so as to change the sequencing of the list.

	p (in my version) starts out as pointing to the 'head' variable,
	so if the search ends there, assigning to *p is assigning to head.
	If the search continues, p ends up pointing to head->next, then
	to head->next->next, etc.  Whenever the search ends, we have a
	pointer to the 'next' value which needs changing.  Except that
	'head' isn't a special case; the pointer might be to a 'next'
	member or it might be to the 'head' variable.