When assured, such a simple datum is for the sake of guarding the act of inference. Color may mean a lot of things; any red may mean a lot of things; such things are ambiguous; they afford unreliable evidence or signs. To get the color down to the last touch of possible discrimination is to limit its range of testimony; ideally, it is to secure a voice which says but one thing and says that unmistakably. Its simplicity is not identical with isolation, but with _specified_ relationship. Thus the hard "facts," the brute data, the simple qualities or ideas, the sense elements of traditional and of contemporary logic, get placed and identified within the art of controlling inference. The allied terms "self-evident," "sensory truths," "simple apprehensions" have their meanings unambiguously determined in this same context; while apart from it they are the source of all kinds of error. They are no longer notions to conjure with. They express the last results attainable by present physical methods of discriminative analysis employed in the search for dependable data for inference. Improve the physical means of experimentation, improve the microscope or the registering apparatus or the chemical reagent, and they may be replaced tomorrow by new, simple apprehensions of simple and ultimate data. _b_) Natural or spontaneous inference depends very largely upon the habits of the individual in whom inferring takes place. These habits depend in turn very largely upon the customs of the social group in which he has been brought up. An eclipse suggests very different things according to the rites, ceremonies, legends, traditions, etc., of the group to which the spectator belongs. The average layman in a civilized group may have no more personal science than an Australian Bushman, but the legends which determine his reactions are different. His inference is better, neither because of superior intellectual capacity, nor because of more careful personal methods of knowing, but because his instruction has been superior. The instruction of a scientific inquirer in the best scientific knowledge of his day is just as much a part of the control (or art) of inference as is the technique of observational analysis which he uses. As the bulk of prior ascertainments increases, the tendency is to identify this stock of learning, this store of achieved truth, with knowledge. There is no objection to this identification save as it leads the logician or epistemologist to ignore that which _made_ it "knowledge" (that which gives it a right to the title), and as a consequence to fall into two errors: one, overlooking its function in the guidance and handling of future inferences; the other, confusing the mere act of reference to what is known (known so far as it has accrued from prior tested inquiries) with knowing. To remind myself of what is known as to the topic with which I am dealing is an indispensable performance, but to