These excellent figures would be difficult to improve upon.

1.

I completely disassembled the watch, cleaned it, and lubricated it
to factory specifications. The only deviation from factory lubrication
entailed the substitution of Moebius 9020 (synthetic) for Moebius 8141
(a non-synthetic). The mainspring barrel was lubricated with Glissalube
20 on the barrel walls and Moebius 8201 on the mainspring
itself, and on the barrel floor and top. No intentional adjustments
were made to the escapement, although removal of a balance cock from
a movement necessarily puts some forces on the balance spring and regulator.
Following the service, regulation with the Triovis adjuster and movable
stud carrier was carried out. As show at left, gross adjustments
of rate are made by moving the regulator itself (red arrows).
Very fine rate adjustments are made by turning the Triovis screw (at
the orange arrow), which moves the regulator in very fine increments.
Beat is adjusted by moving the stud carrier (blue arrows), which
shifts the outer attachment of the spring. Following service, regulation
of rate and beat produced the following figures.

It is noteworthy that these significantly different figures are a result
of nothing more than disassembly, cleaning, lubrication, and reassembly.
Although relative amplitude (between horizontal and vertical) remains
approximately in original proportions, overall amplitude is higher,
probably due to changes in the barrel lubrication which allowed more
tension at full wind (less slip of the bridle). Cleaning and lubrication
of the gear train also generally improve amplitude because of decreased
friction and improved power transfer to the escapement. The spread between
dial-up and crown-down amplitude, 30 degrees, is now three times the
original figure (more typical, probably due to improved lubrication
of the pivot tips). Overall adjustment, however, is not as good as the
original figures. There is now a nine second spread from fastest to
slowest positions and this spread is, unfortunately, between the “260″ height=”238″ align=”RIGHT” border=”2″ naturalsizeflag=”3″ alt=”Engraving on barrle bridge”>two
most critical positions, dial up and crown down. There is also a six
second difference between dial up and dial down positions.

2.

The dial up/down discrepancy required correction before any other adjustments
could be carried out. Because there was little difference in amplitude
between the two positions (eight degrees), I suspected the rate difference
was more likely due to non-parallelism of the regulator index than a
fault in the balance pivots or their lubrication. I thus made a very
slight adjustment to the regulator index (opening it at the bottom),
demagnetized the movement, and reregulated the watch for rate and beat.
This produced similar figures, but the dial up/down discrepancy was,
in fact, worsened.

3.

At this point, there was enough amplitude difference between the two
positions, that the lower dial-down amplitude might be playing at least
some role in the faster rate. But a 16 degree amplitude difference
is probably not enough, by itself, for a 9 second difference. Were amplitude
the issue, the balance pivots, pivot lubrication, and KIF shock absorbers
would have to be checked for irregularities and differences between
the two pivots. Dial up, the balance is supported largely on the upper
pivot tip, dial down on the lower.

The faster dial down rate (without a larger amplitude difference) suggested
that the balance index was still tighter at the bottom than
at the the top. Although misalignment was not visible at 35 power magnification,
I very slightly opened the regulator space at the bottom still further
(blue arrow), and also shifted the stud in the stud carrier up
slightly (red arrows). Although a shift would cause the spring
to be a bit out of flat, the shift was so slight (.1 mm or so) that
any tilt introduced was imperceptible. The change in stud height did
allow the spring to ride a touch higher in the regulator (green arrow).
The results, however, were not promising:

The amplitude difference persisted and seemed increasingly the cause
of the dial-up/dial-down rate difference. I decided to let the watch
run overnight and “settle in,” rewound it fully in the morning and found
the following results on the timer.

The
over-all slight decrease in amplitude (and, thus, increase in rate)
was expected, for a slight deterioration in amplitude is typical of
freshly service watches as lubricants distribute themselves. The dial-up/dial-down
spread of eight seconds remained, but now there was a substantial spread
of 15 seconds between dial-up and crown-up (it had earlier been in the
2-9 second range). At this point it was clearly time for some more work
on the escapement.

4.

I lowered the stud in the stud carrier to its original position, as
this had solved nothing (and put the spring slightly, if imperceptibly,
out of flat). I then cleaned and reoiled the shock-jewel assembly on
the upper pivot. I might have redone the lower pivot,
because dial down
the balance rides on the lower pivot, and the amplitude was lowest dial
down. But, the original oil drop on the upper pivot was a bit on the
small side (about one-third the diameter of the cap jewel instead of
the one-half preferred), and the size of the drop has some relation
to the endurance of the lubrication. I felt that this smaller oil drop
might have reduced the drag enough to produce the amplitude difference
I observed. In other words, I was trying to equalize the drag on the
upper and lower pivots by increasing the drag on the upper pivot, which
is not normally the desirable approach. The reoiled upper pivot is shown
above left, the yellow arrow indicating the edge of the
circular oil drop. Following this work I reregulated rate and beat,
yielding the following figures:

5.

The
amplitude difference between dial-up and dial-down was, indeed, reduced
a bit (from 19 to 8 degrees); and the rates were closer, within 5 seconds.
The critical dial-up/crown-down difference, however, had increased from
11 seconds to 14 seconds. Crown-left and crown-up maintained approximately
the same spread from dial-up. At this point, horizontal centering of
the spring seemed the likely cause for the remaining differences, and
I made extremely small adjustments to both the regulator sweep and dog
leg, both with the balance installed in the watch. (This can be difficult
because the balance cock obscures some of the spring.) I then reregulated
for rate and beat, yielding the following figures:

6.

The situation between dial-up and dial-down had been worsened again,
but it was now largely the dial-up position that was out of line with
the other figures. All other figures are within 8 seconds and the improved
horizontal centering had brought the vertical positions
into better adjustment. (A large change in rate with little change in
amplitude implicates centering.) The dial-down amplitude suddenly seemed
inexplicably on the low side, but I decided to ignore it for the time
being. The very slow dial-up rate suggested that, in adjusting the regulator
sweep, I had inadvertently opened up the index at the top. As adjustment
of the sweep involves reshaping and bending the spring, the delicate
brass index (with a diameter of about 0.10 millimeters) is easily distorted.
I tightened the index high, and spread it very slightly low (blue
arrows, above right). Without any reregulation, the figures
were: