344 PROCEEDINGS OF THE AMERICAN ACADEMY. 



level of the centre of the field of view ; it might be two or three milli- 

 meters too high or too low. It would not do to readjust the telescope 

 for each ball ; and therefore it was highly important that the micrometer 

 cross-hair, which in one part of the work was set on the plumb-line and 

 in another part was set on the tip of the plug, should be strictly vertical. 

 This condition was readily attained during the observations on the plumb- 

 line, with which the work of each evening began ; but the jar of the balls 

 dropping into the pans had a tendency to loosen for a moment the fasten- 

 ing of the telescope, and during this instant the unbalanced weight of the 

 micrometer screw, etc., sometimes caused a slight rotation of the instru- 

 ment, with corresponding inclination of the cross-hair from the vertical. 

 The full measure of the danger of error from this quarter was not real- 

 ized before April 23 ; but after that date, and especially after May 23d, 

 much care was given to the telescopes, and their performance was, on 

 the whole, satisfactory. 



It has already been intimated that the tallow mixture used in the re- 

 ceiving pans gave more or less trouble. At the temperature about 15° 

 C, which prevailed in the base of the tower during the earlier weeks of 

 the work, pure tallow was rather too hard, crumbling too much or crack- 

 ing under the impact of the balls. To remedy this defect lard was added. 

 After some weeks the temperature in this part of the tower rose a degree 

 or two, and then the mixture of tallow and lard was too soft, so that the 

 balls fell through it and struck the bottom of the pan. 



Then beeswax, added to the tallow and lard, was used to give the right 

 degree of hardness. The necessity of such adjustments was a little 

 troublesome, but there were more serious difficulties. No mixture which 

 was used appeared to be homogeneous when cold. All were more or 

 less mottled, whiter in some places than in others, and the whiter spots 

 appeared to be softer than the others. This conditiou of the arresting 

 medium probably accounts in some measure for the irregularities of posi- 

 tion of the arrested balls, the mixture giving way more readily on one 

 side of a ball than on the other side. Such irregularities, however, must 

 in the long run come to naught, if the variations of hardness are purely 

 accidental in their distribution. They prolong the research, but do not 

 affect the final result. Variation of hardness with distance from the edge 

 of the pan has already (in a preceding paper) been mentioned, and the 

 means taken for eliminating, in some measure, error from this source have 

 been described. Such error would apparently not affect the southerly 

 deviation of the balls ; but it probably does affect to some extent even 

 the final result along the east and west line ; for the behavior of the tal- 



