Lawless order (Q1059651)

R. Baer asked (c 1960) if every linearly ordered group can be given a new group action with respect to which it is an Abelian linearly ordered group. The answer is easily seen to be yes if the linearly ordered group is countable since, as a linearly ordered set, it is an ordinal number of copies of the integers ordered in the natural way or the product of the rationals and such a set ordered lexicographically. In contrast, the authors prove that there is a linear ordering \(\leq\) of the free group \((F,\cdot)\) on \(2^{\aleph_ 0}\) generators such that if * is any group action on the set F with respect to which \((F,*,\leq)\) is a linearly ordered group, then the smallest variety of groups containing \((F,*)\) is the variety of all groups. The construction of the ordering is by transfinite induction. The key lemma needed to show that it satisfies the conclusions is: Let \((G,\cdot)\) be a group free on an uncountable set of generators which contains a countable subset dense in \((G,\leq)\). Suppose that every \(\phi \in Aut<G,\leq >\) is piecewise definable (i.e., for every interval I of the Dedekind completion \((\bar G,\leq)\) of \((G,\leq)\), there is a subinterval \(I'\neq \emptyset\) of I such that for some \(a,b\in G\), \(x\phi =axb\) for all \(x\in I'\cap G)\). Then if \((G,*,\leq)\) is a linearly ordered group, \((G,*)\) generates the variety of all groups. That every member of Aut\((F,\leq)\) is piecewise linear for the constructed linear order \(\leq\) on the free group \((F,\cdot)\) uses an ultrafilter game. There is a consistency result about the number of ''lawless'' orders and the existence of such of cardinality less than the continuum. The same technique has been used by the authors to show the existence of a right linearly ordered group such that no new group operation on the linearly ordered set makes it into a linearly ordered group. [Total orders whose carried groups satisfy no laws, Proc. 1st International Conference on Ordered Algebras, Luminy 1984 (to appear)]. The proof is certainly one of the most ingenious in the entire subject; its only drawback is that is doesn't seem to provide a technique for problems other than those of a similar ilk to Baer's.