HALL. — DEVIATIONS OF FALLING BODIES. 341 



considerable irregularity marked their behavior to the very end of the 

 investigation. This irregularity was probably due to a variety of causes, 

 among which the most obvious were, initial sidewise motion, due to 

 vibrations of the tower, and variations of hardness in the arresting 

 mixture. The method of release used, or, rather, the method of fixing 

 and controlling the ball just before release, is evidently one which makes 

 the ball share an}' accidental slight movements of the top of the tower, 

 such as are maintained almost constantly during the day, and much of 

 the time at night, by disturbances coming through the ground from ma- 

 chinery or street-cars or athletic exercises in the vicinity of the labora- 

 tory. Difficulty from this cause was reduced by working at night, and 

 by releasing the ball at an instant when a mercury surface, in a glass 

 vessel resting on the top of the box a (Fig. 1), gave an apparently un- 

 disturbed imasre of crossed threads attached to the side of the vessel. 



At one time, about the middle of April, the experiment of releasing 

 the ball from a condition of free suspension just beneath the upper plate 

 was tried. This method required sharp scrutiny of the ball or of the 

 suspending thread to determine whether pendulum movements of serious 

 magnitude existed before release, and was therefore very trying to the 

 nerves, besides being comparatively slow. Moreover, the pendulum 

 movements never did quite die out, and the performance of balls dropped 

 from a condition of free suspension seemed no better, on the whole, than 

 that of balls dropped from the condition of suspension controlled by 

 contact. Before April 26 I abandoned the method of free suspension. 



April 18, I tried another experiment, using one of the bell-metal 

 balls as a counterpoise for the similar ball to be dropped, and making 

 the contact of the latter with the plate as light as it could well be made. 

 This method of procedure also was tedious aud not particularly success- 

 ful, so that it was soon given up. Once or twice afterward, on May 5, 

 for example, one of the balls was used as a counterpoise, no unusual 

 attention being paid to the character of the contact, which of course 

 must have been rather light; but I presently came to the conclusion that 

 better results were obtained by means of the usual counterpoise, which 

 weighed 93 gm., about 20 gm. more than the weight of a ball. This 

 held the ball firmly in place till the thread was burned, and though the 

 pressure was doubtless in many cases considerably greater on one side of 

 the circle of contact than on the other, any lateral thrust from this 

 source after the burning of the thread must have lasted an exceedingly 

 short time, too short for any harm. The method of release to which I 

 finally settled down, aud which has now been sufficiently described, has 



