And laughter will be more pronounced still, if we find on the stage not merely two characters, as in the example from Pascal, but several, nay, as great a number as possible, the image of one another, who come and go, dance and gesticulate together, simultaneously striking the same attitudes and tossing their arms about in the same manner.  This time, we distinctly think of marionettes.  Invisible threads seem to us to be joining arms to arms, legs to legs, each muscle in one face to its fellow-muscle in the other:  by reason of the absolute uniformity which prevails, the very litheness of the bodies seems to stiffen as we gaze, and the actors themselves seem transformed into automata.  Such, at least, appears to be the artifice underlying this somewhat obvious form of amusement.  I daresay the performers have never read Pascal, but what they do is merely to realise to the full the suggestions contained in Pascal’s words.  If, as is undoubtedly the case, laughter is caused in the second instance by the hallucination of a mechanical effect, it must already have been so, though in more subtle fashion, in the first.

Continuing along this path, we dimly perceive the increasingly important and far-reaching consequences of the law we have just stated.  We faintly catch still more fugitive glimpses of mechanical effects, glimpses suggested by man’s complex actions, no longer merely by his gestures.  We instinctively feel that the usual devices of comedy, the periodical repetition of a word or a scene, the systematic inversion of the parts, the geometrical development of a farcical misunderstanding, and many other stage contrivances, must derive their comic force from the same source,—­the art of the playwright probably consisting in setting before us an obvious clockwork arrangement of human events, while carefully preserving an outward aspect of probability and thereby retaining something of the suppleness of life.  But we must not forestall results which will be duly disclosed in the course of our analysis.

V

Before going further, let us halt a moment and glance around.  As we hinted at the outset of this study, it would be idle to attempt to derive every comic effect from one simple formula.  The formula exists well enough in a certain sense, but its development does not follow a straightforward course.  What I mean is that the process of deduction ought from time to time to stop and study certain culminating effects, and that these effects each appear as models round which new effects resembling them take their places in a circle.  These latter are not deductions from the formula, but are comic through their relationship with those that are.  To quote Pascal again, I see no objection, at this stage, to defining the process by the curve which that geometrician studied under the name of roulette or cycloid,—­the curve traced by a point in the circumference of a wheel