Sperry Gyrocompass

Enclose the rotor before running it, balance it carefully, and run up slowly. A flywheel that leaves its bearings at speed is dangerous out of all proportion to its size.

Sperry claims an electrically driven gyro wheel with a float supporting it in stable equilibrium. The suspension is as much of the invention as the gyroscope.

Put a compass on a table and bring a steel bar near it. The needle swings. Now imagine the bar is the ship — that is deviation, and it changes with every heading.

Mount a heavy balanced wheel in gimbals and spin it hard. Try to tilt the axis and feel it resist — angular momentum wants to stay pointing where it points.

Apply a torque and watch the axis move at RIGHT ANGLES to your push, not along it. Precession is counter-intuitive and it is the mechanism the whole instrument uses.

A gyroscope in full gimbals holds a fixed direction in space and appears to drift as the Earth turns beneath it. That is a directional gyro, and it is not a compass.

Remove one degree of freedom so the axis cannot tip freely. This constraint is what converts a direction-holder into a north-seeker.

Add a weight underneath, or use a mercury reservoir as Sperry did. Gravity now produces a torque whenever the axis tilts — and the tilt is produced by the Earth turning.

Earth's rotation tilts the axis; the pendulous weight resists; the resulting torque precesses the axis horizontally. The precession only stops when the axis lies north-south, where the tilting vanishes.

Carry the mass in a liquid so bearings take almost no load. Friction is the enemy: a stiff bearing produces a torque indistinguishable from the signal and the compass settles on the wrong heading.

Run it and watch the axis oscillate slowly about north, swinging less each time. That long approach is why a gyrocompass takes hours to settle and is left running.

Introduce a small offset in the pendulous torque. The path spirals in rather than swinging forever — undamped, the instrument oscillates about north indefinitely and never gives a reading.

Repeat the step 3 test on your gyrocompass. It does not move. This is the entire point: it is immune to every steel plate and every motor on the ship.

The settling torque depends on the horizontal component of Earth's rotation, which falls with latitude and vanishes at the poles. A gyrocompass is weakest exactly where a magnetic compass is worst.

The patent. US 1,242,065, "Ship's gyroscopic-compass set", filed 25 September 1909 and granted 2 October 1917 to Elmer A. Sperry. The gap between filing and grant is unusually long and worth noting, because Sperry is routinely credited with "the 1908 gyrocompass" — that refers to his working system, not to this grant. He patented in England first and faced sustained opposition from Hermann Anschütz-Kaempfe, whose German gyrocompass of 1908 preceded his; the resulting litigation is famous partly because Albert Einstein served as an expert witness for Anschütz-Kaempfe. Sperry's device was adopted by the US Navy in 1911.

Why a gyroscope alone is not a compass. A freely gimballed gyro holds its axis fixed relative to the stars, so as the Earth turns it appears to wander — useful as a short-term heading reference, useless as a compass. The conversion requires deliberately REMOVING freedom: constrain the spin axis to the horizontal plane and make the system pendulous. Earth's rotation then tilts the axis, gravity acting on the offset mass produces a torque, and because a gyroscope responds to torque at right angles, that torque precesses the axis in azimuth. The motion only ceases when the axis lies in the meridian — so it settles pointing at the geographic pole.

It finds TRUE north, and that is the practical prize. A magnetic compass points at the magnetic pole, which is in the wrong place and moving, and is deflected by the ship's own iron and electrical machinery. Deviation had to be measured for every vessel and corrected with compensating magnets, and it changed with heading, latitude and cargo. A gyrocompass has none of this: it references the planet's rotation, so no amount of steel affects it, and it needs no variation correction because it already points true.

Its own weaknesses, honestly. The settling torque depends on the horizontal component of Earth's rotation, which is proportional to the cosine of latitude — so a gyrocompass grows sluggish and finally useless above roughly 75-80 degrees, the region where magnetic compasses also fail. It must be running for hours before it is trustworthy, and it needs damping or it oscillates about north forever. It also requires speed and latitude corrections, because a moving ship adds its own component to the apparent rotation. Sperry's company went on to build autopilots, bombsights and inertial navigation from the same principles, and merged with Burroughs — whose calculating machine is elsewhere in this batch — to form Unisys.