We cannot rely, however, upon the natural tendency of the observer, because his attitude is likely to change as the experiment proceeds, and a change of attitude means a disturbance of the experimental conditions. Moreover, there are observers of intermediate tendency, who accent _both_ the ‘perception of stimulus’ _and_ the ‘reaction as soon as,’ and may accent them in different degree. Hence it is necessary to instruct the observers at the outset that they are to perform either a =sensory= or a =motor= reaction, that is, that they are to look forward either to the perception of the stimulus or to the execution of the movement. With this preliminary instruction, the sensory reaction takes, on the average and for practised reactors, a tenth of a second longer than the motor, whether the stimulus be a sight, a sound, or a touch. The longer time points, of course, to a more complicated nervous path; and that in turn raises the presumption of a richer mental accompaniment. Observations show, in fact, that _only the sensory reaction represents a complete impulsive action_; the motor reaction does not fall under our formula. The main difference—and we have no space for detailed analysis—is this. The instruction for the =motor= reaction sets up kinæsthetic sensations of strain in the reacting member, principally in the finger; these are contextual processes (p. 118), which carry the meaning ‘You are to react as quickly as possible’; and they are accordingly known as ‘sensations of intended movement.’ They imply that the instruction is already in part fulfilled; the muscles are, from the very first, prepared for the movement that shall end the experiment. Indeed, an observer who is thus instructed will sometimes react prematurely, before the stimulus has appeared, and is also liable to accept as _the_ stimulus any chance stimulus that intervenes, and so to react wrongly. The instruction for the =sensory= reaction, on the other hand, sets up an expectation of the stimulus; the organism is thus prepared especially for perception; premature and wrong reactions do not occur. The intention to move is present, to be sure, but it is in the background, carried only by the feel of the finger as it lies upon the key, or in more diffuse form by the feel of the extended arm upon the table. We might therefore say that, in the motor reaction, the formula tends to close up on itself, like a telescope; idea of result is always approaching perception of result, and intention of movement is always approaching perception of movement; the perception of object gets squeezed between the two extremes, as these draw together; whereas, in the sensory reaction, the formula is followed _in extenso_; the mental processes are thinned out, as we have put it, but they are all present, following one another in their regular order. The reaction experiment thus renders the impulsive action manageable, puts it at our