SECTION 2

TWEAKING THE MARK XII: PART 2.3

ADJUSTING THE MARK XII/887

7.

The dial-up rate did, as expected, increase more than dial-down, but the spread between the two was still too large. Now, the dial-up amplitude was also dangerously high (threatening knocking with any activity), but the previously low dial-down amplitude problem seemed to have spontaneously corrected. Changeable factors too small to see even under very high magnification--including debris in pivots, differences in lubrication, or tiny, imperceptible changes in spring and regulator geometry--often play a role in precise adjustment. The variability of too many measured parameters now suggested that the escapement needed some examination and reworking. I removed the balance cock, balance and spring from the watch. An examination of the lower balance pivot revealed no visible defects at 35 power magnification, but I used a pivot polishing tool and a bit of jewelers rouge on a piece of pegwood, to polish the pivot. I expected this to improve dial-down amplitude. With the balance spring out of the watch, the outer coil (before the regulator sweep) looked a bit out of center, and I reshaped that, as well as the regulator sweep (which was repositioned by the adjustment of the outer coil). I expected the horizontal centering to improve positional performance. Then, having reinstalled the balance cock in the movement, I reregulated for rate and beat on the timer, and the figures were finally promising:

	DIAL UP	DIAL DOWN	CROWN DOWN	CROWN LEFT	CROWN UP	(CROWN RIGHT)
AMPLITUDE	306	302	265	281	286	269
RATE	-2	-2	+2	+1	-1	+2
BEAT ERROR	0.2	0	0.1	0.1	0.1	0.2

The amplitude figures were close to or better than original factory figures. The maximum rate difference in five positions was now four seconds, although that spread fell between dial-up and crown-down. Factory adjustment showed only half that (two seconds) on these two critical positions. My crown-left adjustment, the second most important vertical position (sitting with your arm on the arm of a chair), showed a three second difference from dial-up, identical to the factory's. My crown-up was within one second of dial-up, compared to the factory's zero difference. This was not a bad adjustment, but it was not as elegant as the factory's adjustment, despite the marginally smaller total spread. It did not represent much of an improvement, if any.

8.

The factory point of attachment of the balance spring at the collet (see "Tweaking the Mark XII, Part 2.2"), which I had not changed, should have provided the fastest vertical rates in the crown-left and crown-up positions. My last reading showed the opposite of these expected figures, with slightly higher rates in the crown-down and crown-right positions. I thus suspected that there was still some horizontal centering error in the spring and very slightly adjusted the regulator sweep and dog leg to attempt improvement on the centering. Because the regulator sweep must remain stationary to accommodate the regulator, any adjustments made to it or the dog leg displace the entire rest of the spring. The illustration right, shows how adjustments of the spring at 1, 2, and 3 cause displacements at 4 and 5.

These very, very slight adjustments, and a reregulation of rate and beat, finally brought the watch into an adjustment that clearly improved on the original factory figures:

	DIAL UP	DIAL DOWN	CROWN DOWN	CROWN LEFT	CROWN UP	(CROWN RIGHT)
AMPLITUDE	311	293	281	279	282	273
RATE	+1	0	+1	+1	-1	+2
BEAT ERROR	0	0.1	0.1	0.2	0.1	0.1

All adjusted positions are with two seconds of each other, and, most significantly, the three "important" positions (dial-up, crown-down, and crown-left) are identical. In the ever-quirky way of micro-mechanics, dial-down amplitude was still lower than dial-up, but this was no longer producing a higher dial-down rate--probably because of changes in centering of the spring. Amplitude in the three adjusted vertical positions was essentially identical, with the (more expected) result that these positions were very close in rate. The crown-up rate suggested that there was still some minor residual horizontal decentering of the balance spring, but there always is. One would not advisedly attempt to correct for such a small error. It would be too easy to introduce much larger errors.

9.

It now remained to check adjustments at partial states of wind, and to check rate in actual daily use. I placed the Mark XII, fully wound, on a Cyclomatic-S timer, which holds the watch at a 45 degree angle, thus splitting the difference between dial-up and various vertical positions. But I removed the rotor to prevent automatic winding. After 22 hours, the watched had gained two seconds and showed the following timer readings:

	DIAL UP	DIAL DOWN	CROWN DOWN	CROWN LEFT	CROWN UP	(CROWN RIGHT)
AMPLITUDE	277	273	255
RATE	+10	+12	+13
BEAT ERROR	0.1	0.3	0.3

Lower amplitude figures (from lower mainspring tension) were clearly evident. And, with lower amplitude, came the expected faster rate--although these figures seemed a bit too high. Beat had also drifted some. The spread between dial-up and crown-down was now three seconds. Any residual faults in the escapement are always more influential with lower amplitude because irregularities have a greater influence on a shorter swing. It was also likely that rate had "settled in" and need a little reregulation. So I fully wound the watch before any adjustments to compare it to the last fully wound test:

	DIAL UP	DIAL DOWN	CROWN DOWN	CROWN LEFT	CROWN UP	(CROWN RIGHT)
AMPLITUDE	306	306	291
RATE	+4	+6	+5
BEAT ERROR	0.2	0.3	0.2

10.

Even at full wind, the rate had picked up a bit, and beat was still slightly off. Reregulation of rate and beat yielded:

	DIAL UP	DIAL DOWN	CROWN DOWN	CROWN LEFT	CROWN UP	(CROWN RIGHT)
AMPLITUDE	301	305	283	279	280	283
RATE	+1	+1	+4	+3	+2	+2
BEAT ERROR	0.1	0.1	0	0.1	0	0.2

11.

Over the next 24 hours, I wore the watch for 12 hours, during which it lost 2 seconds. It sat dial up for 12 hours, during which it gained 0.5 seconds. The total 24 hour rate in use produced a loss of about 1.5 seconds.

12.

At the end of the second day, 12 hours in wear, 12 hours dial-up, the watch had gained one second and stood at a total loss of 0.5 seconds.

13.

The next 24 hours was spent back on the Cyclomatic and produced a gain of 2.5 seconds, for a total gain of 2 seconds for the three day period.

14.

The Mark XII spent the next three days on the winder, and gained 32 seconds during this time. So I decided to look at it on the timer again:

	DIAL UP	DIAL DOWN	CROWN DOWN	CROWN LEFT	CROWN UP	(CROWN RIGHT)
AMPLITUDE	311	300	246	289	281	285
RATE	+6	+8	+11	+9	+8	+7
BEAT ERROR	0	0.2	0	0.2	0.2	0

The crown-down position was once again showing somewhat low relative amplitude and subsequent higher rate (now 5 seconds from dial-up). And the crown down position was, once again, too fast. In the crown- down position the weight of the spring places more pressure on the index (rather than the boot) and I wondered if some irregularity in the index might be binding the spring, making it effectively shorter, and increasing rate.

15.

I thus burnished the inside surface of the index with a very fine, polished steel needle, and slightly opened the spread between index and boot (to slow the vertical positions slightly). Because the watch had spent it's three days on the winder in a safe, I also demagnetized it with the case back still off.

	DIAL UP	DIAL DOWN	CROWN DOWN	CROWN LEFT	CROWN UP	(CROWN RIGHT)
AMPLITUDE	302	305	284	285	279	280
RATE	+1/+3	+1	0	+4	+2	+2
BEAT ERROR	0.2	0	0	0.1	0.1	0.2

On this reading, the evenness of amplitude across horizontal and across vertical positions is remarkably good. The dial-up rate was slightly unstable, however, and the timer clearly showed why. As shown at right, there was now some chatter of the spring dial-up, due to some irregular interaction between spring and regulator.

16.

I subsequently adjusted the regulator index to close it slightly, and made very slight adjustments to assure that the regulator sweep was as perfectly vertical as I could make it. If the balance spring is not absolutely vertical as it passes between index and boot, one edge of the spring can contact one of the regulator parts and produce chatter and uneven rate. I then backed off the Triovis screw to slightly reduce the overall rate. These provided my final figures in adjustment of the Mark XII/887. They may be compared to the original factory figures below the first table.

FINAL	DIAL UP	DIAL DOWN	CROWN DOWN	CROWN LEFT	CROWN UP	(CROWN RIGHT)
AMPLITUDE	313	303	285	293	292	291
RATE	+1	0	+2	+3	+1	+3
BEAT ERROR	0.1	0.2	0.2	0.3	0.2	0.1

FACTORY	DIAL UP	DIAL DOWN	CROWN DOWN	CROWN LEFT	CROWN UP	(CROWN RIGHT)
AMPLITUDE	292	288	282	263	273	275
RATE	+7	+5	+5	+10	+7	+6
BEAT ERROR	0.2	0	0	0.1	0.1	0.2

DISCUSSION

In daily use, my final figures provide a watch that runs within a second or two a day, generally losing a second or so during the day (full wind, high amplitude) and picking it back up at night (partial wind, lower amplitude). Occasionally checking against a reference, the watch is plus or minus one second, with no longer-term drift in either direction. As I rarely wear a single watch for more than a week or two, I do not have impressions beyond this period.

The improvement over original factory figures is a small one in terms of utility, but substantial in terms of watch adjustment. My adjustments achieved a total spread over the five positions of three seconds (versus the factory's five seconds); a dial-up / crown-down discrepancy of one second (versus the factory's two seconds); and a dial-up / crown-down / crown-left discrepancy of two seconds (versus the factory's five seconds). Including the unadjusted crown-right, my adjustments provided a spread over the six positions of three seconds, compared to the factory's five.

Is such "tweaking" worth the considerable work involved? That all depends on why you're doing it. As an exercise in watch craft--and sometimes witchcraft--it can be very satisfying. As a practical matter, it is certainly not worth the cost. The precision of the adjustments I made provided a watch that--so far--hardly ever needs resetting. But following the next cleaning, it will have to be done all over again. And, in the meantime, a good hard knock may well undo much of my work by forcing a distorting excursion of the balance spring.