The optimism about the possibilities of DNA computing is based on two centr
al issues: the Watson-Crick complementarity and the massive parallelism of
DNA strands. While the latter issue renders exhaustive searches possible an
d thus may settle problems previously considered intractable, the former is
sue is the cause behind the universality of many models of DNA computing. M
oreover, complementarity can be viewed as a purely language-theoretic opera
tion: undesirable circumstances in a string trigger a transition to the com
plementary string. This aspect of complementarity is investigated in the pr
esent paper, mainly from the point of view of L systems. New types of word
sequences will be discovered. Sometimes the resulting decision problems are
equivalent to well-known open problems from other areas. (C) 2001 Elsevier
Science B.V. All rights reserved.