shot, is sufficient to counterbalance this advantage ; so 

 that whil^ the source of error in Mr. Griffiths' experi- 

 ments is the width of the cluster, by the other method 

 uncertciinty is caused by the length of the group of shot. 



In our own experiments we have found it preferable 

 to sub^itute a rigid steel plate, with a light spring contadl 

 at the back, for the wire screen. The spring may either 

 be made so light as to break the circuit on the impad of 

 a single pellet, or by a certain amount of stiffening three 

 or four shots arriving in rapid succession may be required 

 to interrupt the current. In any case, the spring is made 

 sufficiently stiff and the plate sufficiently heavy to prevent 

 the sound wave from breaking the circuit. 



If two of these plates are arranged at different dis- 

 tances, so that each receives adjacent patches at the 

 centre of the pattern, the time taken for the shot to pass 

 from one plate to another may be measured with con- 

 siderable accuracy up to 40 yards ; serious error could 

 only be caused by one side of the pattern travelling faster 

 than the other, which is most unlikely to occur, unless 

 there is balling or clustering, which is easily detected. 

 By this arrangement striking velocities at any required 

 distance may be taken simultaneously with the velocity 

 from the muzzle to ranges of five or ten yards, at one of 

 which points a few fine wires are cut by the shot, whil^ 

 the latter is still in a compad mass. 



It is possible thus to study the loss of velocity by 

 various sizes of shot under different conditions. 



In some recent experiments published by the "Field," 

 an attempt is made to imitate the well known experi- 

 ments of Bashforth, in which a single heavy projectile 



49 



