Experiments itpon Gunpowder. 59 



accidental ; but as the difference in the velocities, com- 

 puted by these different methods, was in no instance 

 considerable, not being in any case so great as what fre- 

 quently occurred in the most careful repetition of the 

 same experiment, and as the velocities, as determined 

 by the recoil, were much more regular than those shewn 

 by the pendulum, as appears by comparing the curves, 

 g, /, and m y #, (Fig. 16) with the crooked line c, d^ I 

 think we may fairly conclude that this new method may 

 with safety be relied on in practice. 



The greatest difference in the velocities, as ascertained 

 by the two methods, appears in the instance of the 85th 

 experiment, where the velocity, determined from the re- 

 coil, exceeds that shewn by the pendulum, by 346 feet 

 in a second, the former velocity being that of 2109 

 feet in a second, the latter only 1763 feet in a second; 

 and in the two succeeding experiments, the velocities 

 shewn by the pendulum are likewise deficient, though 

 not in so great a degree. 



This apparent deficiency remains now to be accounted 

 for; and first, it cannot be supposed that it arose from 

 any imperfection in Mr. Robins's method of determin- 

 ing the velocities of bullets, for that method is founded 

 upon such principles as leave no room to doubt of its 

 accuracy ; and the practical errors that occur in making 

 the experiments, and which cannot be entirely prevented, 

 or exactly compensated, are in general so small that the 

 difference of the velocities in question cannot be attrib- 

 uted to them. It is true the effect of those errors is 

 more likely to appear in experiments made under such 

 circumstances as those under which the experiments we 

 are now speaking of were made, than in any other case; 

 for the bullets being very light, the arc of the ascent of 



