Quasigroups, loops, and associative laws (Q1814982)

The author investigates the question of which weakenings of the associative law imply that a quasigroup is a loop. In particular, he completely settles the question for all laws which are written with four variables, three of which are distinct (``size four laws''). In earlier work [J. Algebra 183, No. 1, 231-234 (1996; Zbl 0855.20056)], he had shown that every quasigroup satisfying any one of the four Moufang laws is a loop. In the paper under review he provides now a wealth of information on size four laws, culminating in theorems stating that, other than laws which are equivalent to the full associative law, each of the following laws and their ``mirrors'' (obtained by writing the equation backwards), and each of the Moufang laws, are those whose validity implies that a quasigroup is a loop: \[ \text{(i) }(x(xy))z=(xx)(yz),\quad\text{(ii) }(x(yx))z=(xy)(xz),\quad\text{(iii) }(x(yy))z=x((yy)z), \] \[ \text{(iv) }(x(yy))z=(xy)(yz),\quad\text{(v) }(x(yz))y=(xy)(zy),\quad\text{(vi) }((xy)z)x=x(y(zx)). \] According to the author, his investigations have been aided by the automated deduction tools OTTER, FINDER, and MACE.