THE ROYAL ARTILLERY INSTITUTION, 
51 
inaccurate than one fired at a shorter range, even though the real accuracy of 
the first gun might be the greater of the two. 
15. But there is a consideration not yet alluded to which has an 
important bearing on the point now under discussion. 
In describing the case above investigated, we assumed at the outset, for 
the sake of simplicity, that the gun had no tendency to err more in one 
direction than another, and therefore deposited its shots in a circular form 
round the central point of impact, as shewn in Eig. 
Now this is very seldom the case, for both theory and experience shew 
that there is almost always a greater tendency to err in one direction than in 
another. It is true that many causes of error tend equally in all directions ; 
but there are two important ones, almost always present in artillery 
practice, which act in one direction only. The first is, the variation in 
the initial velocity of the shot, caused chiefly by irregularities in the strength 
or burning of the powder, and which tends to produce vertical error on the 
normal target, or what is called “ error in range ” on the ground. The 
second is the disturbance of the position of the gun by the commencement 
of its recoil, which also acts in the same direction as the former one, the first 
movement of the axis being almost exclusively in elevation. 
These two causes combining together will tend to magnify the vertical 
errors on the normal target, and so to produce ail elongated cluster somewhat 
like this figure. 
Eig. 3. 
The proportions of the elongation will be very variable. Indeed, I have 
met with some cases in which the figure, contrary to the usual rule, has been 
elongated in the horizontal instead of the vertical direction; but these are 
exceptional, the probability being that if enough rounds were taken, we 
should always find the vertical exceed the horizontal dimension.* 
16. This fact, that the errors of the shot follow a different law in the 
vertical and horizontal directions, naturally points to another method of 
estimating their positions, far more simple and appropriate than by taking 
their absolute distances from the centre. This is the plan generally followed 
* This peculiarity of figure in tlie practice of guns deserves more investigation. It might be 
easily inferred from tbe probable rectangles, if the angle of descent were ascertained; or the 
adoption of the “ test target ” system which I propose hereafter, would shew it as a part of the 
process itself, without trouble. I am informed by Captain Noble that in some very carefully made 
experiments with a 12-pr. Armstrong gun, the variation in the “ angle of departure ” was about 
five minutes. 
