342 PROCEEDINGS OF THE AMERICAN ACADEMY. 



the merit of being very expeditious, the twenty-four balls constituting 

 an evening's tale being, as a rule, dropped in less than one hour and a 

 half. 



At the receiving end also of the apparatus there was considerable ex- 

 perimenting. At first the pans were not fastened in the supporting seat 

 in such a way as to preclude the possibility of rotation under the shock 

 of the balls, and it is quite possible that slight rotations did occur, some- 

 times in one direction and sometimes in the other, their net effect being 

 presumably zero. A locking device for preventing accidental rotation 

 was first used April 7 ; but until May 21, the screw then adopted as a 

 part of the lock in Fig. 2, was not employed, a wedge being inserted in 

 the lock beside the stud when it appeared to be needed. 



Previous to April 16, but not later, the receiving pan (see Figs. 4 and 5), 

 after catching ball No. 1, was turned counter-clockwise to take position 

 for ball No. 2, in the same direction for ball No. 3, and so on. This 

 practice dropped the 1st and also the 6th ball into a field symmetrical as 

 to its north and south aspects ; but it dropped the 2d, 3d, 4th, and 5th 

 balls into a field having a ball already lodged in the southerly quarter, 

 with no corresponding ball in the northerly quarter. The average clear 

 distance between neighboring balls in the pan was probably rather more 

 than 3.5 cm., and it seemed unlikely that, as a rule, the lodgment of a 

 ball in the tallow would be appreciably affected by the lodgment of an 

 earlier ball 3.5 cm. away ; but in some cases the tallow was inclined to 

 crumble for a considerable distance about each ball, and in such cases 

 there was sometimes a crumbling all the way through from a ball to its 

 nearest neighbor. Evidently the practice of turning the pan always in 

 such a direction as to make the majority of the balls fall to the north of 

 their nearest predecessors might make the crumbling more prevalent on 

 the south than on the north of the balls at the critical instant of lodg- 

 ment; and therefore the displacement of the balls due to this crumbling 

 might show a balance toward the south. To avoid this possibility the 

 practice was adopted of rotating the pan first in one direction and then 

 in the other, so that, after April 16, the order of arrangement of the 

 six balls in any pan corresponded to the order of arrangement of the six 

 studs projecting from the pan, as numbered in Fig. 5. 



Too little attention was paid, at first, to the telescopes with which 

 observations on the plumb-line and on the point above the balls (see Fig. 

 4) were made. As the depth to which any ball would sink in the tallow 

 was somewhat uncertain, it was not possible to adjust the telescopes in 

 advance so that the tip of the plug above the ball would fall just at the 



