NavList:

Bill Morris you wrote: Thank you, David, foryour comments.

people might assume that the gyro artificial horizon is the panacea, and that it must be preferable to the bubble or the pendulous reference artificial horizons. What do aircraft pilots prefer? A piece of wool taped to the winscreen, a turn and slip indicator or an artificial horizon indicator?

I might have caused some confusion here. When I said artificial horizon, I meant with respect to sextants, not the aircraft flight instrument.  I assume you chose your example more to indicate good, better, best, but even then, it’s not the best analogue to bubble, pendulous reference, gyro, or pendulous reference, bubble, gyro, whichever way you choose to put it.

Wool taped to the windscreen shows the direction of the outside airflow and will indicate if the aircraft is side-slipping.  The turn and slip indicator is actually two separate instruments inside a single case.  The slip ball is effectively an upside down spirit level and it fulfils a similar function to the wool taped to the windscreen.  The turn indicator is a rate gyro with a horizontal, not vertical, axis any turning about the aircraft’s yaw (Z) space axis causes the gyro to attempt to precess about its longitudinal roll (X) axis.  However, this is restrained by springs, so that the rate of turn is indicated.  You can’t really topple a turn indicator, so it’s the pilots No1 friend when recovering from unusual positions and spins, particularly inverted spins when, even with a visual horizon, the pilot’s eye-brain linkage tends to be easily confused.

The gyro artificial horizon fight instrument is a pendulously mounted gimballed gyro with a near vertical axis.  It’s certainly subject to acceleration errors and is not difficult to topple if subjected to unusual positions.  It can be erected using a rapid erect button which effectively realigns the gimbals to the frame.  Vacuum driven instruments are also kept vertical from any wander by little pendulous vanes which half cover vents in the front, back, left, and right sides of the gyro inner case.  If an aircraft accelerates forwards, the pendulosity of the casing causes the horizon bar to show the aircraft is flying right wing low, and the pendulosity of the side vanes cause the horizon bar to show the aircraft to appear to be safely climbing.  In the late 40s and early 50s when aircraft performance was increasing rapidly, this error is deemed to have accounted for a number of accidents occurring on dark nights to high performance aircraft, which crashed shortly after take-off and to the left of the runway centreline.

You also wrote:
then there must be a side force applied to the pivot by the cup, and the gyro will begin to precess. I don't doubt this, but it must be for a very short time, and of very small magnitude, as the following suggests. I trained the sextant on the sun near noon, when its altitude was changing slowly. Rapidly rocking and moving back and forth, swaying and moving from side to side. and rotating resulted often in loss of the sun's image from between the graticule lines, but as soon as I got it back between the lines, the graticule embraced it perfectly steadily.

That’s more or less what I just found with my MkIX bubble sextant looking at Sirius, and that’s where regular time gap averaging and faithful following comes in with the IXAs and IXBs.  Dickie Richardson, and probably Hughes before him, realised that a steady bubble didn’t necessarily mean zero acceleration; it just meant constant acceleration, whatever the value.  But I’m really an aircraft navigator, and the accelerations of an aircraft are usually more even and of a longer period than those of a yacht.  My only knowledge of submarines is from watching Das Boot.  

Then you wrote:

Might some of this be due to hang-off caused by Earth rotation? I am not certain that I have completely understood you, but I am inclined to doubt this, as I used a horizontal collimator set with a ten second level, which of course senses local vertical.

I’m afraid I don’t know anything about collimators so I can’t really comment except to say that when you have something apparently drifting off at a constant rate, e.g. a gyro trying to be fixed in space enclosed within an Earthbound frame, and that gyro is being corrected by an Earth related correcting force, there’s bound to be a hang off error, because you can’t develop a correcting force before drift has occurred.  Eventually a balance is achieved resulting in a hang-off error. However, I can find no mention using Google of 'hang-off error', so I’m either completely wrong, Google's too young to have heard about it, or I’ve invented a new term. DaveP