My recent attempt is with a Norland, 3-holer. Last was with a Norland 2-holer - still not going the distance (even after re-bushing the barrels!). So I need to go back to first principles, starting with the springs. I consulted my references - the most learned company I can think of - Gazely, deCarle etc.
Heard of Robinson? Well here's some wise words written in 1934 ... "By far the greatest number of mechanical clocks derive their motive power from springs. The modern high grade mainspring forms a very reliable and compact power storage component, and when housed in a properly designed barrel, provides an excellent driving unit for a timekeeper".
Unfortunately, there is sometimes the mistaken idea that mainsprings and barrels having plenty of power, and operating energetically under almost all conditions, require little attention, and that, unless there has been a definite failure at this point, nothing more than a lookover is necessary.
To hold this view is as fallacious as it would be for an automobile engineer to argue that because the gear box and back axle of a car were in perfect order, the engine needed no attention. It is useless to have a train whose depths and pivots are perfect, and whose escapement or chiming and striking mechanism are correctly adjusted, unless the motive power is efficient.
To perform this work and to keep doing it over a long period of years, the spring and barrel must be as mechanically perfect as possible. Only by careful design and construction can the correct result be obtained. The ideal conditions are when the mainspring supplies just enough power to drive the train smoothly and evenly throughout the run.
Roughness of construction may be overcome by fitting stronger springs than those theoretically necessary. This is often done in badly made clocks, but such measures are basically wrong. They impose heavier loads on barrel arbors, pivots, teeth and clickwork, for no better purpose than to make good errors and rough workmanship elsewhere.
The horologist should take pains to see that the springs and barrels of the clocks which pass through his hands are in such order as to give their best performance. Springs for replacements should be of the highest quality, and of the same height and strength as those originally in the barrel. Nor should fracture alone be the reason for replacement. Few horologists seem to renew a spring for any other reason than its breakage, but it should not be forgotten that springs can fatigue, after years, and that this will result in a serious loss of power.
The mainspring is perhaps the most severely used spring in the whole of mechanical engineering practice. The service demanded of it is extremely arduous, so that even the best mainsprings lose something of their elasticity under continual use. A spring of the best quality, and of correct height and thickness, will retain its elasticity for a very long period, but even a good spring of wrong size will have a short lived efficiency, and will fatigue or fracture in a comparatively little while. More than this, with such a spring the regulation of a clock to anything like proper timekeeping is impossible. (Doesn't this make the absolute sense, even today 70 years on?)
deCarle describes how to select a new spring ... "When selecting a new spring, there are one or two points to consider. The maximum number of turns are obtained if the diameter of the barrel arbor is one third of that of the interior of the barrel and the mainspring occupies one third with one third space. If this condition is observed the outer coil of the spring when fully wound will occupy the position of the inner coil when the spring is run down. The number of turns the barrel will make equals the difference between the number of coils when fully wound up and run down. For instance, if there are 15 coils when run down and 25 when fully wound up round the arbor, the barrel will make 10 turns : 25 - 15 = 10."
But the best guide in my opinion is Conover, "Another problem mainspring which should be replaced is a "tired" or weak spring". It is difficult to define exactly what is meant by these terms, however ...
The mainspring on the left is an old one removed from a barrel. The mainspring on the right is new. The old mainspring on the left has a much reduced capability, as shown by the fact that when fully released it is not much larger in diameter than a barrel. In contrast, the spring on the right is new and has never been in a barrel before. It expands to a large diameter. This is not to say that good springs will always look like this one when they are removed from barrels for cleaning and inspection. I have verified that a new spring will be somewhere between the extremes shown in the figure if it is removed from the clock after a test period. Experience will teach you to recognise which springs are weak and which ones are not. One of the most pointed lessons is learned by cleaning and reusing an old spring which looks questionable, only to have it run a clock for six days instead of eight. The entire movement may have to be dismantled to permit the replacement of the spring with a new one. Some repairers replace almost every mainspring in the clocks they repair. Although that is not my policy, I can understand why some people approach mainsprings that way.
Ordering a Hole End Mainspring
Selecting a replacement barreled mainspring is sometimes as easy as measuring the old one. As a convenience for repairers, most supply catalogs list mainspring dimensions in English and metric units. I prefer to use a micrometer to measure the thickness and a rule to measure the width in millimeters.
Thickness of a mainspring must be correct within close limits. A mainspring of 0.3mm thickness is a relatively weak spring, the type used in small French movements. By increasing the thickness by just 0.15mm, we arrive at the powerful springs used in much larger movements. Thickness becomes a compromise when you cannot find the exact mainspring you need in suppliers' catalogs. My approach is to avoid using a stronger mainspring than the one which came out of the clock unless I know it was too weak a spring. A spring of 0.45mm thickness can wear out the gears in an old movement if the correct spring was 0.4mm. Backing up this general approach is the idea that today's mainsprings have more torque than 100 year old mainsprings of equal thickness had when they were new.
Width of a mainspring is a little less critical than thickness. If the only available mainspring of the correct thickness and length is 1 mm narrower than the original spring, this does not usually present a problem. A wider mainspring cannot be used because it will not allow the barrel cover to fit securely, and in addition, there will be no endshake, or freedom, in the fit of the barrel arbor.
Length of an old mainspring can be measured directly if you try to stretch it out, although there is some estimating involved because of the tight inner coils. You can also measure the inner diameter of the barrel and compare it to the "diameter" figure included in suppliers' mainspring tables. This figure represents the coiled diameter of the wire bound mainsprings as they are shipped to you. If the coiled diameter fits in your barrel, it is close to the correct length.
It is far better to use the formula below to find the correct length mainspring for your barrel. This gives you an exact length for the spring of your chosen thickness that will turn the barrel the maximum number of times. This is an important concept because a mainspring which is too short will obviously give a reduced number of barrel turns. It is equally true that a mainspring which is too long will not give the maximum number of possible turns because it fills up some of the empty space in the barrel required for winding and unwinding. The formula is based on the idea that a mainspring occupies only a part of the available space in the barrel. For a barrel to be able to deliver the maximum number of turns possible with a mainspring of a specified thickness, the spring should occupy one half the available area.