5 16 DEDUCTIONS FROM SOME FIELD ARTILLERY PRACTICE. 
(c) That from o% to 25% on graze gives good results for scoring 
“ men hit.” 
(d) That from 4% to 40% gives good effect for “ hits.” 
(e) That there may be good effect without any grazes, that, con¬ 
sequently, it is not always possible to base corrections of fuze on the 
proportions of “ Airs ” to “ Grazes.” This is why recourse must be had 
to the proportion of “ highs ” to “ lows,” for the actual heights of burst 
can seldom be correctly estimated. The judging of “ lows ” of course 
introduces a personal error, but not, I think, a very variable one. 
Fig. X. shows: — 
(a) That there should be no + grazes for the best average effect. 
(b) That for good effect there should be few. This is why re¬ 
course must be had to “ + lows” (which are at any rate more numerous 
than + grazes) when we wish to correct our elevation during Time 
Shrapnel fire. 
(c) That to count in time with percussion Shrapnel grazes when 
analysing, will bring out too high a gun-range. This is evident, for 
there should be no + grazes. 
(d) How difficult it must be to range with Time Shrapnel by 
trying to preserve the proper proportion of + to — grazes, or 
even + “ lows ” to — grazes, and how careful it is necessary to be 
before making corrections of elevation upwards during Time Shrapnel 
fire. Only about 10% should be + grazes even with a total of 50% 
on graze. This proportion varies, but not directly, with the total per¬ 
centage on graze. The moral seems to be, adequate verification be¬ 
fore Time Shrapnel. But all this can only be seen from many such 
figures as Fig. X. 
Fig. XI. shows that “men hit” are a much less pronounced measure of 
general accuracy than “ hits.” As “ hits ” alone would leave distribu¬ 
tion out of account; a curve of “ hits ” x “ men hit ” has been drawn. 
Though it is not in direct proportion to accuracy, it measures it very 
fairly. If the target is larger or the number of rounds fewer, Fig. 1 
has shown us that the measure would be a good deal fairer. In any 
case all the figures have shewn that “ hits ” are a better measure than 
“ men hit ” ; it seems unwise therefore not to reckon in both for score. 
Fig. XII. is similar to Fig. VI. but about three times as many rounds 
have been taken into account. About 32 rounds per series were fired 
at different targets. The conclusions are the same as those drawn 
from Fig. VI. 
Fig. XIII. shows : — 
(a) How important it is to reject any round about whose observa¬ 
tion there is the least doubt. One round misobserved in five would 
seem to reduce the effect by 50%. We have already seen that several 
extra verifying rounds would not seem to entail any very evil results, 
the repetition of doubtful rounds is therefore emphasised. 
(b) How the range-finding seems to have influenced the result 
when observation was bad is seen by the very fair effect when 40% to 
50% were misobserved. 
(c) That the loss in effect due to misobservation is serious: the 
