In previous articles about fly development, I’d gone from the maternal gradient to genes that are expressed in alternating stripes (pair-rule genes), and mentioned some genes (the segment polarity genes) that are expressed in every segment. The end result is the development of a segmented animal: one made up of a repeated series of morphological modules, all the same.
Building an animal with repeated elements like that is a wonderfully versatile strategy for making an organism larger without making it too much more complicated, but it’s not the whole story. Just repeating the same bits over and over again is a way to make a generic wormlike thing—a tapeworm, for instance—but even tapeworms may need to specialize certain individual segments for specific functions. At its simplest, it may be necessary to modify one end for feeding, and the opposite end for mating. So now, in addition to staking out the tissues of the embryo as belonging to discrete segments, we also need a mechanism that says “build mouthparts here (and not everywhere)”, and “put genitalia here (not over there)”.