Having switched the original Mark XII caliber 884
for a caliber 887 (see “Tweaking the Mark XII:
Part 1“), I spent in excess of 50 hours
tweaking the escapement of the 887 to see if I could improve on the
already excellent performance of this movement as delivered from the
factory. In Part Two of the Mark XII/887 story I will discuss the timing
and adjustment of watches in considerable detail. I have thus divided
the discussion into three parts. This first part, 2.1, is about the
concept and meaning of watch timing and adjustment; part 2.2 will elaborate
on the the functioning of the escapement and the technical concepts
used in adjusting a watch (see also, “The Escapement” in the Horologium);
and part 2.3 will detail the long–and sometimes arduous–history of
adjustments carried out on the Mark XII/887 and the results obtained.

ABOUT ACCURACY

There is no subject that, even among sophisticated
watch collectors, is as misunderstood as timing adjustments to a watch.
People routinely speak of “accuracy” as if it were a unified concept,
and judge a watch heavily on the criterion. The term is usually used
to mean simply that the watch compares favorably to a time standard
in whatever unspecified way the owner happens to have compared it. When
a watch does not compare
well–adjusted at the factory with no idea of how the ultimate owner
will use it, shipped from Switzerland half way around the world, and
stored for months or years in a dealer’s showcase–the owner is disappointed
and deems the watch “inaccurate.” The idea that one watch is “more accurate”
than another because it gains two seconds a day, while the other gains
eight, is a naive and simplistic approach to judging the quality–and
the timing–of a watch. By this standard, unadjusted watches often
appear to perform better than fully adjusted watches, which by any informed
understanding they rarely do.

WHAT IS REALLY BEING MEASURED

“Accuracy” is used to imprecisely describe a group
of four, essentially separate, parameters–parameters that, when
taken together, determine the ability of a watch to match a time
standard with relative consistency in specified kinds of use. An understanding
of the real parameters behind the popular idea of accuracy will allow
a finer appreciation of watches, help the collector judge the quality
of a watch,
and give him a sense of what is necessary to correct a problem with
timing should it occur. These four parameters are my own construction
and are not traditionally described in the literature on timing
and adjustment. They are, however, useful concepts and underlie all
timing and adjusting procedures.

The first of the four parameters behind the idea of
accuracy is stability of rate in a single position (SRSP). One
would like the rate of a watch in, for example, a (stationary) dial-up
position to be be as stable as possible. But no mechanical watch meets
this criterion absolutely, and the rate of very good watches fluctuates
constantly. The SRSP of a very finely made watch may deviate over a
range of four to eight seconds (calculated as deviation from absolutely
correct rate, per day). The deviation of a less well made–or less well
set-up–watch may span as much as 20 seconds or more. On an electronic
timer, the rate of any mechanical watch–calculated and displayed
as seconds plus or minus per day–fluctuates continuously. But it is
important to emphasize that the SRSP parameter has, in itself, no relation
to “correct” rate. SRSP is simply about stability. SRSP is a
measure of the (relatively) instantaneous stability of rate,
with no regard to a time standard. Assuming no damage and good servicing,
SRSP is largely a product of the refinement and quality of the escapement
and
particularly the escape wheel, pallets, and pallet lever. Without reasonably
good SRSP, adjustment to positions and regulation of absolute rate are
moot. SRSP is an adjustment issue (as opposed to a regulation
issue).

The second parameter underlying accuracy is the stability
of averaged rate in a single position (SARSP). If we average
the rate of the watch over one five minute period, how close is this
average rate to the average rate of any another five minute period?
The second period may be measured consecutively with the first, or days
later. We hope that the average rates will be relatively close. But
as anyone who has used an electronic watch timer knows, watches are
surprisingly variable–and sometimes absolutely quirky–even in this
relatively easy specification. The averaging of rate over longer periods–but
not long enough that state of wind becomes a factor–reduces
the effects of uneven running reflected in the instantaneous fluctuations
described by SRSP. But the “averaging-out of errors” that occurs is
never perfect and over finite amounts of time involves a considerable
element of chance. In other words, in practice, averaging all error
together never produces a perfect result, even if theoretically it might.
And any measured performance is only a sample that suggests
performance over longer periods. In addition to the quality and condition
of the escapement, the SARSP can be easily affected by the quality and
condition of the entire movement, all the way back to the mainspring.
SARSP is an adjustment issue (as opposed to a regulation issue).

The third parameter underlying the accuracy of a watch
is the relative averaged rate in different positions (RARDP).
This parameter necessarily includes SARSP (because measurements are
necessarily made over time), with the addition that rates are compared
not only between time periods, but between different positions.
For the wristwatch, these traditionally include the two horizontal and
three (of the possible four) vertical positions. For example, the averaged
rate (SARSP) of the watch dial-up (DU) over a five minute period is
compared to the averaged rate crown down (PD, for “pendant down”) over
five minutes. RARDP is, of course, the performance in question when
we speak of a watch being “adjusted to positions.” Averaged rate between
time periods (SARSP) is also necessarily being measured in this specification
(because a watch cannot be measured in two different positions simultaneously).
If the SARSP is poor, positional performance cannot really be determined.
RARDP is dependent on the refinement and condition of the escapement,
the repair and condition of the entire gear train of the watch, and
the very fine adjustments applied by the timer (meaning watchmaker
responsible for timing) to the balance wheel, balance spring, balance
spring collet, balance pivots, and regulator pins (if any). RARDP is
the best predictor of the potential for the watch being regulated
to maintain a rate consistent with a time reference in typical daily
wear. RARDP is an adjustment issue (as opposed to a regulation
issue).

Finally, the fourth and simplest parameter underlying
the idea of accuracy is simply absolute rate (AR). In this context,
absolute refers to the time keeping of the watch relative to
a known time standard. Without good stability over time and between
positions, absolute rate is a moot point because it cannot be reliably
accomplished. With good stability over time and between positions,
absolute rate is a simple matter of adjusting the effective length of
the balance spring with the regulator (or, in the case of adjustable
mass balances, the center of mass of the balance). Although absolute
rate is the parameter most noticed by the watch owner–and it is, no
doubt, important–it is the simplest to accomplish and is the end-product
of other, more complicated parameters. For the watchmaker, a “rate adjustment”
is a simple matter. AR is a regulation issue, as opposed to the
adjustment issue pertinent to the first three parameters.

ADJUSTMENT TO POSITIONS

Watches that are “adjusted to positions”–usually
five, but sometimes three, four, or six–are considerably more expensive
than unadjusted watches. This is at least partially justified because
of the time put into adjustments. But there is a relationship between
cost and the number of adjustments because, for good reason,
adjustments are usually done only on better movements. A well-made,
expensive movement may or may not be adjusted; but a poorly made or
inexpensive movement cannot be adjusted. But, despite their considerable
extra costs, it is often observed that in daily use adjusted watches
often seem to run at no better an absolute rate than many unadjusted
watches.

The
adjustment of watches–and final regulation for performance in terms
of absolute rate–has everything to do with how they will and will not
be used. All wristwatches of reasonably high quality can be adjusted
to provide relatively consistent performance in one position
(good stability of averaged rate in a single position). Because
of this consistency, they can subsequently be regulated to provide relatively
good absolute rate (AR) in a single position. Such an “unadjusted” watch
is thus, in fact, adjusted and subsequently regulated in one position,
usually dial up. Today, machine-manufactured and timed balance-spring-staff
assemblies make this quite easy, even in relatively inexpensive watches.
Forty years ago this was not true and inexpensive watches did not run
as consistently, even in a single position, as they do today.

A watch “adjusted to a single positions” will provide
a good absolute rate so long as it is used in the position it is adjusted
to, or to the extent that errors in other positions fortuitously
average each other out. When such averaging
happens, the absolute rate may be quite good, despite significant inconsistencies
of running. Many people who report good absolute rate for their watches
are not aware that this is being accomplished by the averaging of relatively
poor rates at different times and in different positions. Were the watch
checked at different times of the day, it might be found to vary significantly.

By timing a strap wristwatch in two positions–dial
up and crown down (PD)–a manufacturer stands a very good chance of
providing a watch that appears to be “accurate.” Such two-position timing
is also relatively easy with current technology. Placed dial-up at night–which
can amount to nearly 50 percent of the running time of the watch–and
hanging crown down during the day whenever the wearer is walking with
arms hanging, the watch is likely to put in a good averaged performance.
If by chance, activities vary on a particular day, the inconsistent
rate is simply chalked up to chance. A doctor once told me that his
faithful (and unadjusted) Omega watch “ran slow” only when he
attended continuing-medical education
classes. He attributed this error to his impatience and boredom with
the class, but wondered if the watch could be corrected anyway. During
classes, he was not only bored, but sitting uncharacteristically with
his arm on a table for eight hours, the watch held largely in a crown
right position (as seen from the movement side of the watch). Were he
a regular desk worker, he would have long ago found that his watch “ran
slow” and had the rate adjusted to account for the crown right position.
In this case, he might have been complaining about the rate of his watch
only on the rare day he spent away from his desk with the watch crown
down.

The adjustment of a watch to five or six positions
is intended to maintain the absolute rate of the watch with greater
immunity to the variability of actual use. In other words, a fully adjusted
watch is intended to run more consistently regardless of how it used.
It is more independent of position than a watch adjusted to one or two
positions. This does not mean, as some expect, that regulation of absolute
rate will be perfect from the factory for any particular owner.
The full adjustment of a watch means simply that it is capable
of such regulation because it can run relatively consistently in five
or six positions. Because adjustments can never achieve identical
rates in all five or six positions, individual habits will still affect
the absolute rate of any mechanical watch.

CONCLUSIONS

In addition to the obvious pragmatic reasons for fully
adjusting a watch, adjustment is an important–perhaps primary–component
of watch craft. Even in the context of current technology, there is
no watchmaker so exalted as the expert timer, and no watch so fine and
refined as that fully adjusted to positions, temperature, and isochronism.
The fully adjusted mechanical watch often seems infused
with magic that defies the physics that would otherwise dictate its
mortal limitations.

Throughout the last two centuries, the number of adjustments
applied to a watch has been accepted as a primary measure of its quality.
There is, no doubt, much reason behind this understanding, for the very
fact that a watch can be adjusted in five positions is a measure
of its quality. But there is a very important caveat here for the collector
of contemporary watches. A watch can be “adjusted to positions” to any
desired tolerances. Would it be correct to say that a watch adjusted
to five positions and within 30 seconds per day from the slowest to
the fastest position is fully adjusted? Probably not. In recent years,
I believe some Swiss manufacturers have used assumptions about the traditional
meaning of adjusted to positions to market watches that do not merit
the designation.

In Part 2.2
of “Tweaking the Mark XII,” I will discuss the technical concepts underlying
adjustments to a contemporary wristwatch like the Mark XII. In
Part
2.3 I will discuss the actual history of adjustments to the Mark
XII/887 in pursuit of possible improvement on the already excellent
factory adjustments.