upon Gun-powder^ i&ro* 
Recoil 
Weight of 
the ballet. 
Aftual. 
Computed. 
Difference. 
9 2 cl exp. 
2352 
3 2 > 2 5 
3 2 > 2 5 
91 
*754 
27,18 
27,22 
+ 0,04 
90th 
1 184 
21,92 
GO 
Vn 
rs 
O 
0 
1 
89th 
603 
1 5 ’ 33 . 
+ 0,2 
88 th 
'600 
•15,22 
15,29 
+ 0,07 
87 th 
354 
11,03 
11,87 
+ 0,84 
86 th 
25 a 
9,62 
4 0,21 
+ 0 >59 
‘85 th 
90 
j,i 6 
7 > 02 
-0,14 
84th and 93d 
0 
4?4 
At 4 
Here ‘the agreement of the aflual and computed recoils is as 
remarkable as that of the actual and computed velocities in the 
foregoing table. 
By the figure 17. may be feen atone view the refult of all 
thefe experiments and computations. The numbers upon the 
line AB (as in the fig. 16.) reprefent the weights of the bullets,, 
while the lines drawn from thofe numbers perpendicular to AB 
•on each fide, and ending at the curves m, », are as the velocities 
of the bullets in the feveral experiments ; the line AB being the 
axis of the curves, the lengths taken from A to the different 
numbers towards B ( ~ x) the abfciffa, and the perpendiculars 
(=y) the correfponding ordinates. The ordinates to the 
the curve hn 9 are as the velocities computed from the theorem 
and the ordinates to the curve/*, n (which is the 
logarithmic curve, as itisj-rry) Ihew the velocities computed 
upon Mr. n obihs’s principles. The curve gf is drawn from 
U u 2 the 
