We know that the retina of our eye contains two basic types of photoreceptor cells, rods for twilight, and cones for daylight vision. The split in two physiological types is paralleled by the split in two types of morphology, already established in 19th century. However, studies of a vast number of vertebrate taxa revealed wide variations in photoreceptor morphology that arose from adaptive radiation in various visual environments. It is often problematic to reliable assign a physiological term “rod” or “cone” to a clearly distinct morphological type of a cell. In 1942, Gordon Walls proposed that during the evolution of vertebrates there occurred multiple changes of visual habitats in various groups, and such changes might lead to a complete loss of an “unnecessary” photoreceptor type. Further changes in visual ecology might result in a restoration of previously lost receptors from the presently available cells. For instance, according the transmutation theory, mammalian rods and cones are secondary inventions and are not homologous to rods and cones of lower vertebrates. Modern molecular data, however, prove evolutionary continuity of rod and cone lines, from cyclostomata to mammalians. Nevertheless, morphological and physiological transmutations as such are real, and demonstrate numerous parallelisms in developing specific rod-like and cone-like biochemical, physiological, and morphological features. Just recently, we begin to understand the causes of repeatedly appearing motives in rod and cone evolution. The present lecture will discuss the role of morphological details of rods and cones in their corresponding functions. Thus this is an attempt to answer a basic question posed by L.A.Orbeli: “Why evolution proceeded this particular way?”