But, in the first place, there is no evidence, nor any probability, that there is anything corresponding to a retina on the outer face of the crystalline cone; and secondly, if there were, it is incredible that, with such an arrangement of the refractive media as exists in the cornea and crystalline cones, rays proceeding from points in the external world should be brought to a focus in correspondingly related points of the surface of the supposed retina. But without this no picture could be formed, and no distinct vision could take place. It is very probable, therefore, that the visual pyramids do not play the part of the simple eyes of the _Vertebrata_, and the only alternative appears to be the adoption of a modification of the theory of _mosaic vision_, propounded many years by Johannes Müller. {123} Each visual pyramid, isolated from its fellows by its coat of pigment, may be supposed, in fact, to play the part of a very narrow straight tube, with blackened walls, one end of which is turned towards the external world, while the other incloses the extremity of one of the nerve fibres. The only light which can reach the latter, under these circumstances, is such as proceeds from points which lie in the direction of a straight line represented by the produced axis of the tubes. [Illustration: FIG. 29.—Diagram showing the course of rays of light from three points _x_, _y_, _z_, through the nine visual rods (supposed to be empty tubes) A–I of a compound eye; _a–i_, the nerve fibres connected with the visual rods.] Suppose A–I to be nine such tubes, _a–i_ the corresponding nerve fibres, and _x_ _y_ _z_ three points from which light proceeds. Then it will be obvious that the only light {124} from _x_ which will excite sensation, will be the ray which traverses B and reaches the nerve-fibre _b_, while that from _y_ will affect only _e_, and that from _x_ only _h_. The result, translated into sensation, will be three points of light on a dark ground, each of which answers to one of the luminous points, and indicates its direction in reference to the eye and its angular distance from the other two.[10] The only modification needed in the original form of the theory of mosaic vision, is the supposition that part, or the whole, of the visual rod, is not merely a passive transmitter of light to the nerve-fibre, but is, itself, in someway concerned in transmuting the mode of motion, light, into that other mode of motion which we call nervous energy. The visual rod is, in fact, to be regarded as the physiological end of the nerve, and the instrument by which the conversion of the one form of motion into the other takes place; just as the auditory hairs are instruments by which the sonorous waves are converted into molecular movements of the substance of the auditory nerves.