H3 is the third of John Harrison's large navigation clocks. He started work on it in 1738 and it was only ready for sea trials 1757. Harrison referred to it as 'my curious third machine', and he hoped that it would be accurate to within 3 or 4 seconds a week.
H3 survives to this day. It can be seen in working order and keeping good time
in the Navigation Room at the Royal Observatory, Greenwich. It descended to
Harrison's grandson who presented it about 1840 to the Royal Astronomical Society and was kept going until about 1865. It was damaged in 1909 by an unskillful attempt to set it going after it had stopped for
want of winding. It was cleaned and repaired by Lieut.- Commander Rupert Gould
and set going in 1931, and has been going very regularly since
Harrison left no drawings of his clocks. Drawings were made by E.J.Dent, who spent four years from 1836 cleaning the three clocks H1, 2 and 3. but did not attempt to restore the clocks to going order. Rupert Gould, who spent many years restoring this and the other Harrison clocks also left many drawings.
H3 is a large clock weighing 27 kg and stands 59 cm high. It contains 753 separate parts. It is characterized by twin interlinked balance wheels with a large coiled spring attached to the upper wheel and it beat seconds. Both are driven by a grasshopper escapement mounted between their shafts. Large roller bearings support the shafts and small caged rollers, judged to be the forerunners of modern roller bearings are to be found elsewhere in the train. A remontoire reset every thirty seconds is fitted and this is driven by a large spring barrel and fusee. There is a locking mechanism between fusee and remontoire, which prevents the latter from unwinding. A maintaining power is fitted to the winding spring. The clock is fitted with a temperature compensating bimetallic strip or curb, which slides on the balance spring.
The Meccano H3
The Balance Wheels
The large interlinked balance wheels are joined by adjustable crossed wires, ensuring that both balances work together without any
play and also allow complete freedom of movement. The range of excursion of the wheels is about 20 degrees and the period of oscillation is one
second. It is a heavy oscillator with a limited range of motion. The balance spring is a large very short very heavy spring of 1 1/2 turns applied to
the arbor of the upper balance. This oscillator was of course designed to replace the
pendulum and to be impervious to a certain degree of clock movement while mounted in gimbels aboard ship.
Each balance shaft is supported on two sets of supports, one in front and the other behind. These
supports are in fact long arms whose bearings are in the clock plates, but the part which supports the balance shafts are parts of
a very small parts of a large circle and the effect is that of a roller bearing. The
arms are cross-linked by springs to balance their weight.
The grasshopper escapement.
There are two grasshopper palates, one above the escape wheel and the other below. Each
is mounted on a carrier attached to a balance shaft. They are each weighted in such a way that their pallets move away from the escape wheel teeth, and are returned
to a tooth by a controlling spring. This spring also undergoes slight bending after
the pallet connects with a tooth and while the escape wheel gives impulse. Once this
is completed, slight counter- rotation of the escape wheel frees the pallet, which recoils because of the way it is weighted, and
its movement is arrested by the spring, which quickly damps any oscillation before returning it to the next tooth which it does
The Escape Wheel.
There are 120 teeth in the escape wheel and they are of narrow pointed profile and angled in the direction of rotation, which is
The remontoire acts every 30 seconds to ensure constant torque on the escape wheel. It
consists of two rotating parts, one of which is the escape wheel itself but free on the shaft.
The other is the input pinion on the escape wheel shaft to which a star wheel is attached
and fixed to the shaft. Two light springs are stretched between the star
wheel and the escape wheel. If the star wheel rotates anti-clockwise, the
springs will tend to drag the escape wheel round as well. The star wheel
is released every 30 seconds, but the escape wheel is prevented from rotating freely by the pallets, so that there is a constant
tension and release of the springs, but the net result is to keep the escape wheel rotating.
One would think that there is a constantly changing torque of the escape wheel, but this is not so due to a unique method
of attaching the springs to the star wheel. The latter in fact consists of two
diametrically opposed curved horns bent in the direction of the escape wheel's rotation. The springs are attached to the tip, but as the springs are stretched, they bend around the horns in such a way
that their point of attachment moves towards to the star wheel centre and so the mechanical advantage of the spring decreases as
the spring tension increases, the two balancing out to produce an even torque on the escape wheel.
To mention that the mainspring is also fitted with a fusee is to state the obvious, that the clock has belt and braces and shows
the lengths to which Harrison went to ensure even torque on the balances.
A method of preventing accidental unwinding of the remontoire spring is provided. This
consists of long rods stretching from the winding spindle to a ratchet wheel on the centre spindle.
In order for the remontoire to function, a signal is required, and this is provided by the escape wheel itself, because attached
to it is a pin wheel with eight pins. Engaging with a pin is a small cupped lever,
which is slowly depressed as the escape wheel rotates. This lever releases a
vane attached to the train of step up gears to which the main spring and finally the star wheel are attached. The star wheel starts to rotate, stretching the remontoire springs. Fixed
on the star wheel is an eight lobed cam, which further depresses the lever and this function is to allow the cupped lever to be
disengaged from the pin wheel. It then springs back by its own weight and into
position for the next pin. Meanwhile, the cam wheel continues to rotate
and eventually allows the lever to return to the resting position and stop the vane from rotating.
A fly attached to the vane arbor prevents excessive speed of rotation and absorbs inertia.
Harrison is known for his gridiron pendulum, consisting of strips of brass and steel which prevented temperature induced changes
of length in the pendulum. A bi-metallic strip is used in H3 (Harrison called
it his thermometer curb), which is mounted transversely and is fixed to the clock plates on the left side, but its right side
embraces the origin of the oscillator spring by two pegs. As the tip of the
bi-metallic strip bends up or down, so the period of oscillation and thus the mean rate can be changed minutely because the
effective length of the spring has been changed. The strip is changeable both for
length and angulation by vernier screw adjustment. The strip can react must
faster than a gridiron to changes of temperature due to its small heat capacity.
H3 was never taken to sea. It did make a short voyage from London to Plymouth where
it was kept on board ship for months. During the making of H3, Harrison
was directing work on a watch, which became H4. Harrison asked the Board of Longitude
to consider the watch 'as being of great service with respect to longitude' and it was this timepiece that eventually won the
H3 was accurate to within 3/4 second per day. There seems to be little doubt
that Harrison's greatest problem with H3 originated in a very short, very stiff balance spring.
Moreover the characteristics of the spring are very un-isochronous - the restoring force is not directly proportional to
the amplitude. The spring is constrained at both ends, causing it to develop
eccentrically and the forces required to produce this are over and above the winding and unwinding forces. The shorter and stiffer the spring, the worse the effect and the spring is
none linear which will prevent the spring from producing isochronism.
Michael Adler - September 2004