On free products of bands (Q757604)

In a paper of \textit{P. Olin} [Semigroup Forum 35, 265-278 (1987; Zbl 0615.20040)] two normal bands are effectively constructed, the free product of which has an undecidable word problem. The verification of this involves a certain normal form theorem. The author gives a counterexample to show that this theorem is not correct, but then develops some results on the word problem in free products of bands to re-establish the validity of the Olin example. From the existence of this example it is then deduced that subbands of the factors in a free product of bands need not generate a subband isomorphic with their free product.