matophores are "contracted" but the pupil contracts and dilates as before. Therefore the contraction of the iris is independent of the changes in the chromatophores.
At this point physiological experiment had to be abandoned, and it would have been extremely comfortable for Steinach to do as one of his predecessors had done—ride the rest of the way on a cantering hypothesis; but he appealed to histology. In his effort to determine in what other parts of the iris there was pigment, through which the light must produce its effect, he found that his judgment was confused by particles of pigment from the posterior layer, which were scattered at random over his histological preparations. This difficulty was obviated by removing the posterior layer of pigment before making the sections. After taking this precaution he showed that there is no ordinary pigment in the stroma of the iris; neither are there any ordinary smooth muscle fibers like those in the iris of the higher vertebrates. He found the sphincter muscle of the iris composed of spindle-shaped pigmented cells. That these are really muscle fibers he proved by their form, size, characteristic fibrillar structure, and function. It was impossible to observe directly the contraction of these fibers; he adopted the indirect method of killing the iris in the relaxed and in the contracted states and observing the condition of the fibers in each. In the former they were slender and narrow, in the latter shorter and thicker. The ciliary muscle fibers are not pigmented, and this accounts for their being indifferent to the light. His general conclusion is that light produces contraction of the isolated fish and amphibian iris by acting directly on the fibers of the sphincter muscle through their pigment.
The striking characteristic of this investigation is the exhaustive consideration and removal of alternative beliefs. His final conclusion is only an inference, and derives its "certainty" from the fact that it is the only belief that is left. In its relation to this conclusion the evidence is circumstantial. If now the reaction of the pigment and fibers could be directly observed, Steinach's conclusion would be set down as a verified prediction. Though unverified, it is unhesitatingly accepted, like so much of our "knowledge," as an important truth; for most minds its verification would add little or nothing to its certainty, and would even deprive it of some of its interest. This inferential knowledge forms a large part of scientific truth, and other instances of it will appear in the following example of method in morphology.
Various ciliated organs of unknown function in different mollusks had never been brought under the yoke of homology. One of the most decisive tests in morphology for the determination