If these points of attachment were changed to the opposite side of collet so that the spring would develop downward as shown at "G" Fig. 11, and "H" Fig. 12, the results would be reversed and the pendant up rate would be slow in comparison to the pendant down rate.
This point of attachment in which the spring develops downward from the collet is generally known as the slow point among adjusters, and when a spring is pinned at either the slow or fast point the pendant right and left positions generally compare quite closely to each other in timing, provided that the poise and other conditions of the watch are correct.
If the pinning point was changed to the intersection of the collet and vertical line as shown in "I" Fig. 13, the pendant up and down rates would compare nearly equal to each other and the pendant right position would be slow compared to the pendant left position.
If it were pinned at the intersection of the collet and vertical line just opposite to that shown in Fig. 13, the pendant left position would be slow compared to the pendant right position.
The vertical points of attachment are seldom used, for the reason that the variation between the pendant right and left positions would be very difficult to control within close limits, due to the existence of the natural error. As these positions, together with the pendant up position are the most important of the four vertical positions, they are given preference, and the natural error is placed in the pendant down position where it will be the least detrimental to the performance of the watch.
34. _The Natural Position Error and Why it Cannot be Eliminated._
The natural error generally consists of from twelve to fifteen seconds in finely constructed watches, and exists because of the fact that it is impossible to perfectly poise a spiral spring. The location of the heavy point, however, may be shifted by changing the point of attachment at collet as described in No. 33, this Chapter. The nearest approximation of a poised spiral spring is probably attained through L. Lossier's inner terminal curve. Results are not positive, however, and any deviation from the required precision makes the curve valueless. It is possible to obtain perfect adjustment between three vertical quarter positions and the two horizontal positions, but all four quarter positions cannot be perfectly adjusted because the natural error will show up in one of them. Manufacturers of fine watches do not of course presume to supply perfect adjustment in the five positions. Some however, have considerably closer limits of allowance for variation than do others and it is logical to presume that a line of high grade watches having a five position allowance of six seconds from one position to any other would show better results than another line which had even a six position adjustment and an allowance of fifteen seconds from one position to any other.