Generalization of right alternative rings (Q1211105)

Recently, several authors have been examining a generalization of the right alternative law. They assume that a ring satisfies (i) \((x,x,x) =0\) and \[ (wx,y,z) (w,x,[y,z]) =w(x,y,z) (w,y,z)x. \tag{ii} \] In this paper the author adds a third condition: (iii) \((y,y,x)\) is nilpotent for all \(x,y\) in \(R\). He is able to prove: A ring of characteristic not two which satisfies (i)--(iii) and has no proper nil ideals and which has an idempotent \(e\) such that there are no nilpotent elements in \(R_1(e)\) and \(R_0(e)\) must be alternative. The author then correctly concludes that there are no simple rings satisfying the hypotheses. The method is the traditional Peirce decomposition. He shows that the table is right alternative and then alternative and uses the alternative table to show that associators are alternating functions on their entries. It seems on first glance that the author made it easy on himself by assuming (iii) and no nilpotent elements in \(R_1(e)\) and \(R_0(e)\). This appears to make left alternativity easy. Such is not the case; the multiplication of the subspaces is so badly behaved that it is difficult to find elements which are nilpotent and in \(R_1(e)\) or \(R_0(e)\).