124 Professors Perry and Ayrton on the Music 



rods moved in circular grooves ; and the result would be that 

 not only could we alter the amplitude of any one of the com- 

 ponent harmonic vibrations by moving a slider along the 

 rod, but we could also alter the phase by giving the rod a cir- 

 cular motion. To do this, however, satisfactorily would have 

 required either the employment of a much larger roller, so that 

 the slowest vibration of a slider occurred four or five times 

 during one revolution of the roller, or else a change in the 

 arrangement. In the illustration (fig. 1) the glass plate, for 

 simplicity is shown merely kept in position by the four cords ; 

 but in reality it moved in a horizontal frame which again slid 

 in a vertical one, so that any lateral motion at right angles to 

 the plane of the glass was impossible. 



The reason why there is necessarily a considerable dis- 

 tance between the sets of fixed and moved pulleys is in order 

 that a longitudinal motion of the slider shall not alter the mean 

 horizontal or vertical position of the spot on the glass. At 

 first we had the cords much longer than shown in the figure ; 

 but then, even after great care had been taken in endeavouring 

 to obtain inextensible cords, some stretching was found to take 

 place in practice ; consequently we were compelled to deter- 

 mine experimentally what was the greatest length of cord that 

 could be used without _ the stretching interfering with the ac- 

 curacy of the motion ; and this length was the one employed 

 in the actual apparatus. 



The ingenious way in which a number of pulleys are made 

 to give the sum of their motions to the extremity of a cord 

 was suggested to us by the arrangement employed in Sir W. 

 Thomson's tide-calculating machine ; but it is possible that in 

 our new machine we shall adopt a totally different plan, and one 

 which we think is new. If the two extremities of a long rigid 

 rod have parallel motions perpendicular to the rod, the middle 

 of the rod has a motion equal to half the sum of the extremi- 

 ties. Thus the parallel motions of two, four, or 2 n points may 

 be compounded. Similarly for the three points, one third of 

 the sum of parallel motions is obtained from the centre of a 

 rigid triangular piece of which the points are the corners ; so 

 that by bars and frames of simple construction it is easy to 

 get the sum of the parallel motions of any number of pieces. 



We think the roller-plane has much to recommend it; but 

 a series of little cranks may be better. Let a number of par- 

 allel shafts, having their bearings near their ends on two sides 

 of a trough-shaped metal frame, receive from a system of 

 spur-wheels near their centres relative velocities 1, 2, 3, 

 &c. Let A (fig. 13) be either end of any shaft, B being the 

 corresponding bearing ; and let there be attached to it, and 



