To produce movements even of a complicated kind is not difficult.  There exist a great multitude of known contrivances for all the more usual purposes, and if the exertion of moderate power is the end of the mechanism to be contrived, it is possible to construct the whole machine upon paper, and to judge of the proper strength to be given to each part as well as to the framework which supports it, and also of its ultimate effect, long before a single part of it has been executed.  In fact, all the contrivance, and all the improvements, ought first to be represented in the drawings.

320.  On the other hand, there are effects dependent upon physical or chemical properties for the determination of which no drawings will be of any use.  These are the legitimate objects of direct trial.  For example; if the ultimate result of an engine is to be that it shall impress letters on a copperplate by means of steel punches forced into it, all the mechanism by which the punches and the copper are to be moved at stated intervals, and brought into contact, is within the province of drawing, and the machinery may be arranged entirely upon paper.  But a doubt may reasonably spring up, whether the bur that will be raised round the letter, which has been already punched upon the copper, may not interfere with the proper action of the punch for the letter which is to be punched next adjacent to it.  It may also be feared that the effect of punching the second letter, if it be sufficiently near to the first, may distort the form of that first figure.  If neither of these evils should arise, still the bur produced by the punching might be expected to interfere with the goodness of the impression produced by the copperplate; and the plate itself, after having all but its edge covered with figures, might change its form, from the unequal condensation which it must suffer in this process, so as to render it very difficult to take impressions from it at all.  It is impossible by any drawings to solve difficulties such as these, experiment alone can determine their effect.  Such experiments having been made, it is found that if the sides of the steel punch are nearly at right angles to the face of the letter, the bur produced is very inconsiderable; that at the depth which is sufficient for copperplate printing, no distortion of the adjacent letters takes place, although those letters are placed very close to each other; that the small bur which arises may easily be scraped off; and that the copperplate is not distorted by the condensation of the metal in punching, but is perfectly fit to print from, after it has undergone that process.