THE ROYAL ARTILLERY INSTITUTION. 
24. The plan of determining the mean errors from the normal target is 
free from the difficulty experienced with the horizontal target in regard to 
variable descending angles, as the normal position eliminates all error from 
this source. But it is still, like the probable rectangles, inapplicable to the 
comparison of guns fired at different ranges; indeed, I am not aware that any 
system ever yet brought forward has escaped this difficulty. 
25. I venture however now to propose a method which will, I think, 
meet the case, and will at the same time offer some other advantages. It 
is exceedingly simple; and is founded on precisely the same principles as 
those adopted by Captain Noble, being in fact a mere extension of them 
to a form of the problem somewhat more practical, and of more general 
application. 
26. I propose to measure the accuracy of a gun by the probability of Us 
hitting, not an indefinitely small area as hitherto assumed, but an actual 
target , to be called the test target, whose dimensions shall be proportioned 
according to the range at which it is to be used, in a mode hereafter 
explained. 
Eor example: let it be required to compare the accuracy of two guns, 
one firing at 2000 yds. range, the other at 3000 yds. Let the former be 
directed at a target one yard square, and the latter at one larger in the 
proper proportions. Let 1000 rounds be fired from each gun, aim being 
always taken precisely on the centre. Then if one gun hits its target 100 
times (which a good rifled gun will do) while the other hits only 90 times, 
I propose to denote their relative degrees of accuracy by the direct ratios of 
these numbers, or as 10 to 9. 
27. This mode approaches quite near enough to theoretical accuracy* 
while it has the advantage of being a kind of test of the most simple, 
practical and intelligible nature, and one indeed which artillerists and rifle¬ 
men are perfectly familiar with in every day practice. And it has further 
the merit of being applicable to practice at any range. 
28. I proceed to shew, that in order to apply this method, it is not at all 
necessary that the guns should actually be fired at the targets as above 
supposed. It is quite sufficient that the practice be carried on in the usual 
way, by firing the guns at certain elevations, and registering the places of the 
shot as they fall upon the ground, precisely as for the system of probable 
rectangles. All the rest is matter of calculation, as from these data, the 
number of shot which would hit the targets, if the guns were really fired at 
them, can be deduced with even greater correctness than by actual trial* 
The only addition necessary to the usual elements given, is the determination 
of the angle of descent of the shot, which is by no means a difficult matter* 
29. The most convenient form for the test target is a square; it must 
be supposed to be placed normally, and with its centre precisely correspond* 
ing to the centre of impact* 
