NB I am not an aviation professional.

I saw a documentary about the terrible Air France 447 disaster the other day.

I was struck, as were the experts in it, including David Learmount, who's known as a great aviation expert here in the UK, by the way all 3 pilots were completely unable to grasp what was really happening because of their fatal inability to trust the electronic instruments.

I had a thought: is it not the case when flying any aircraft, whatever its size, that you always need to know, particularly in an emergency, two things?

  • air speed
  • attitude

I know this is VERY VERY obvious, but please bear with me.

The fact of the matter is that you can obtain totally reliable information about both these aspects, regardless of darkness, regardless of electronics, regardless of Pitot tubes, at all times:

  • air speed: by having an incredibly robust, small structure outside the front of the cabin window which would flex visibly according to windspeed, thereby making the invisible (air flow) slightly less invisible: at 0 knots this would be inclined downwards; at over 150 knots the windspeed would cause this to become fully horizontal. Lack of horizontality would therefore indicate a low speed. One thing I have appreciated since posting this question is the intensely hostile environment this structure would be facing: freezing temperatures, high windspeeds, sometimes containing ice, etc. No doubt one of the reasons why during the development of aviation this has not been considered possible. But materials are being developed now which are much stronger than ever previously known. One candidate might be carbyne. It would also presumably have to be heated.
  • attitude: by having one of those completely mechanical floating compasses (as used on ships) in a prominent position in the cabin. Much confusion has been created in some of the comments about apparent gravity. But apparent gravity has effect only when acceleration other than gravity (down!) and lift equal to 1G (up!) is present. The fact of the plane accelerating/decelerating (in any or all of the 3 dimensions) could be indicated by a system of calibrated inertial accelerometers in all 3 dimensions, with lights varying in intensity according to acceleration.

Obviously such a bendy thingy would have to be very tough to withstand the speeds, battering, temperatures, etc. But it is at least conceivable that this is not beyond the wit of humanity. Designed carefully, a change in the way it was flexed would then indicate without any possible doubt when an aircraft was in or near a dangerously low speed.

Wouldn't these two measures have had a good chance of providing the pilots of AF 447 with a true and totally irrefutable mental picture of what was actually going on?

PS it has been objected that such a system, even if feasible, would be redundant. But redundancy in critical systems is a desirable thing: what's worrying is if you don't have redundancy, and the aviation industry has made gaffes in the past over insufficient redundancy: Boeing 737 rudder issues; MD80 horizontal stabiliser: "The absence on the McDonnell Douglas MD-80 of a fail-safe mechanism to prevent the catastrophic effects of total acme nut loss" (Alaska Airlines 261, NTSB report quoted).

  • 1
    $\begingroup$ Imagine the drag from a flag at Mach 0.8... $\endgroup$
    – Notts90
    Commented Mar 2, 2017 at 10:23
  • 2
    $\begingroup$ Semi-serious. I would imagine over the life of an aircraft that drag would actually cause a noticeable fuel bill. Any disruption to the airflow has a drag penalty that needs to be justified. Also even at that size it needs to be strong and durable enough to withstand the extremely high wind speeds, which a standard one isn't. $\endgroup$
    – Notts90
    Commented Mar 2, 2017 at 10:30
  • 10
    $\begingroup$ A water compass will not serve as an attitude indicator because apparent gravity inside a plane in coordinated flight is always in the same direction relative to the pilots. $\endgroup$ Commented Mar 2, 2017 at 11:42
  • 3
    $\begingroup$ The external mechanical indicators you suggest would almost certainly be subject to the same icing problems that brought about the errors in the existing instrumentation. $\endgroup$
    – J W
    Commented Mar 2, 2017 at 18:09
  • 2
    $\begingroup$ Ignore the word ‘electronic’ in that report. It's a red herring. They didn't trust the instruments because 1) one failed—in a completely mechanical way—at the beginning of the accident sequence and 2) they were not able to correctly decode the usual combination of indications in the stressful situation. It had nothing to do with the fact the instruments are electronic and everything to do with psychology and some to do with ergonomics of the interface (and electronic interface can be made much more ergonomic—it is just still hard). $\endgroup$
    – Jan Hudec
    Commented Mar 5, 2017 at 20:12

5 Answers 5


Unfortunately, if there had been a jazz band in the cockpit singing "The plane is stalling" it wouldn't have helped. If Moses had been in there carrying a stone tablet newly-carved by God himself inscribed "Pitch down" with a host of angels to help announce it, that would have made no difference.

The unrecoverable problems were in the pilots' brains and in the communication between them, not in the aircraft or the technology.

Once in that state, only a fresh perspective (for example stepping back from the problem, bringing in a fresh person to look at it, using a checklist) is likely to snap those trapped in it out of their spiral: of alarming events, inappropriate reaction, unexpected result of the new inputs, and mounting panic.

It's a common experience in many crafts and disciplines, though in some - like computer programming - it's easier to find the new perspective than in others, such as piloting or surgery.

Anyone who has found themselves in that spiral will be able to tell you that adding more data or new sources of data to the mix rarely help.

It seems reasonable to think that a clear, simple and apparently unarguable piece of evidence would be the way out of such a situation. Unfortunately, in practice, it rarely if ever is, and there is nothing to suggest that in this case it would have been any different - despite compelling intuitions to the contrary.

  • $\begingroup$ The human psychology angle is crucial. But the "semi-reliable mediation" between direct human perception and physical phenomena through the use of electronic indicators, in which the pilots are capable of losing faith seems a huge factor too. I've read stuff out there which suggests that pilots quite often don't believe the stall indicators, just as I don't believe Microsoft's "help" system when it tells me that the only solution to my minor technical glitch is to reinstall every driver or (why not?) my entire system. $\endgroup$ Commented Mar 2, 2017 at 15:52
  • $\begingroup$ According to what I've read (e.g. crankyflier.com/2011/05/31/…) A330s did not, at the time of AF 447 (I don't know about now) have stick shakers. Also the stick was placed to the right of him and he was in the right-hand seat. Do correct me if this is incorrect. $\endgroup$ Commented Mar 3, 2017 at 9:36
  • $\begingroup$ Re stall warnings with pilots placed in a situation similar to AF 447: books.google.co.uk/… $\endgroup$ Commented Mar 3, 2017 at 9:49
  • 7
    $\begingroup$ +1 for the jazz band. The problem was entirely human psychology here, not instrumentation. Here is a video of a wheels-up landing, the pilots are fixated on getting down when they are too high and too fast, notice the screaming noise at 30 seconds in? They didn't, it's the 'gear up' warning. Mike, you need to read up 'human factors' in aviation because you are definitely suffering from task fixation/confirmation bias, oh, and try the gorilla test $\endgroup$ Commented Mar 3, 2017 at 13:54
  • $\begingroup$ One of the findings was that the pilots were commanding conflicting inputs on the stick and were unaware of the other pilot's opposite inputs (which are averaged together by the flight computer). Another is that the stall warning went quiet due to IAS below a certain minimum - contrary to what one might expect. I don't see how one can argue that the problems were entirely in the pilots' heads. $\endgroup$
    – Roman
    Commented Dec 25, 2017 at 0:45

@mike rodent I've read your comments on this post and they seem rather strong in tone to me. I'm joining the discussion here to hopefully explain to you why your ideas are not going to work as you intended in real life.

The reason for the gap between your proposed solution and other's comment on "it will not work" is that you have a flawed understanding of the underlying physics and aerodynamics. That leads to many of your assumptions being wrong.

First thing first, attitude. You proposed:

by having one of those completely mechanical floating compasses (as used on ships) in a prominent position in the cabin

A floating compass can be upright because of gravity. Gravity, when measured as a force, is not absolute; rather, it is relative to your motion. When you're on a roller coaster, you're defining gravity in your own reference frame. When you measure gravity in this scenario, it is impossible to tell if downward acceleration points perpendicularly to the ground. In fact gravity can be pointing up to the sky. You can put tea in a cup and the liquid will never spill, yet you're upside down.

For that reason a mechanical floating device for telling attitude in an airplane is totally useless: you need a gyroscope. It is one instrument the pilots already had and it was functioning on AF 447.

Furthermore, for your information (although it is not related to the question), a mechanical compass will either lead or lag in a turn. You cannot use an ordinary compass to turn an aircraft; if you do, you will never roll out on your desired heading.

For the speed, you proposed:

by having a (fluorescent?) strip/flag, probably metallic, outside the front of the cabin window: at 0 knots this would be inclined downwards; at over 150 knots the windspeed would cause this to become fully horizontal.

What you're describing is not a speed indicator. It does not measure the physical quantity airspeed. Rather, it is an angle of attack indicator, albeit a heavily dampened one.

Outside of the fuel economy and view obstructing aspect, your proposed solution does not really solve the problem you're trying to solve (namely, the pilots do not trust their digital instruments to tell if the plane is above stall speed):

  • An airplane can stall at any airspeed. You can be travelling at 300 knots and stalling at the same time.
  • Stall speed varies when flaps and slats are deployed. Your proposed device cannot be adjusted.
  • Stall speed also varies with the aircraft's load. As you burn fuel, your stall speed is going to change. So does when you load heavy vs load light. Again, your device cannot be adjusted for this change.
  • Temperature variation will change a metal's flexibility. Therefore your device can only be calibrated to work at one ambient temperature.
  • Experienced ambient temperature varies with altitude and more importantly, airspeed!
  • Your device is susceptible to icing.
  • If your device has a large enough surface area, it may also be susceptible to fluttering.

The pilots already had an angle of attack indicator in the Airbus.

The third point that I wish to mention, is that video documentaries are only faithful to the actual event to a certain extent. They are constructed based on actual events, however if one derive an analysis based on a documentary, it can be wrong because the content of the video are not a word-to-word transcript.

Online articles are even worse. In the aviation community we generally disregard any articles, both online and printed, unless it originates from a trusted source tied to aviation, such as AOPA or NTSB.

The only source you should be using when analyzing aircraft incidents is the official report. It contains a transcript of the cockpit voice recorder, which are the exact words spoken and not an interpretation by film makers and actors. It also contain detailed analysis of the scenario, including interview with other pilots, simulations of the exact circumstances with different flight crew in a flight simulator etc.

  • 1
    $\begingroup$ Secondly, I challenge the idea that this is an "angle of attack" indicator. How do you come to that conclusion? Are you assuming that if the aircraft is rising/descending that this would influence the deflection of the strip? But that would depend on its stiffness, surely. Your points about the extreme conditions influencing the ability to deliver useful info are interesting, however. $\endgroup$ Commented Mar 2, 2017 at 13:26
  • 3
    $\begingroup$ @mikerodent yes, that is my point. Rising/sinking would influence the deflection of the strip, as the act of rising/sinking alters the direction of airflow with respect to the airframe (and thus the strip). If you have a drone (or a really strong fan + some wires) you can run your own experiment. You can also ask a separate question either here or on Physics.SE if you are interested in a detailed discussion. $\endgroup$
    – kevin
    Commented Mar 2, 2017 at 13:31
  • 4
    $\begingroup$ when a plane is neither accelerating nor decelerating how do you know?! That's the problem! When all outside references are not available (e.g. night, cloud), how would the pilots know if I should trust the instrument at this instant? And to further on that, yes from a physics perspective, a plane is constantly accelerating even in straight and level flight with constant airspeed. $\endgroup$
    – kevin
    Commented Mar 2, 2017 at 13:39
  • 1
    $\begingroup$ @mikerodent, no, a plane is constantly under acceleration due to gravity. Its velocity vector is constantly changing. If it is not, the plane will float to outer space. $\endgroup$
    – kevin
    Commented Mar 2, 2017 at 16:13
  • 5
    $\begingroup$ "The pilots already had an angle of attack indicator in the Airbus": No, this was a recommendation of the BEA to evaluate the possibility to have a mandatory one. There are AoA probes, the value is used in normal law to protect the flight envelope, but is lost in alternate law. Besides the AoA measures were invalid from time to time. $\endgroup$
    – mins
    Commented Mar 2, 2017 at 19:31

A flag is not really effective because a big jets stalling speed is well over 100+ knots. The flag will be blowing fully even in the stall. I'm not familiar with the ship suggestion but it sounds like you're desribing a gyroscope - aka an attitude indicator, which is already right in front of the pilots.

But furthermore, it doesn't solve the problem. AF447 was a very complicated accident. The plane was literally screaming at the pilots that it was in a stall. All of the indications were there for the pilots to identify that they were in a stall. And yet, the pilots didn't recognise it. Putting another indicator would not do anything to solve this.

As a side point, the third pilot (the captain) did recognise the stall relatively quickly, he just wasn't in the cockpit early enough to help.

EDIT: A metallic flag or something stiff may give some indication that airspeed is low, but it's usefulness is still limited. In a fully-developed stall it would point up, which would only confirm that it is descending, not that it is stalling. It also could dangerously impede pilot visibility, such as when identifying traffic in the terminal area.


Wouldn't these two measures have had a good chance of providing the pilots of AF 447 with a true and totally irrefutable mental picture of what was actually going on?

No more so than the standard indicators of attitude and airspeed that we've had for as long as people have been climbing into flying machines.

Attitude indicator

(source: pilotfriend.com)

Airspeed indicator

But that's just the basic instruments - modern airliners have all sorts of other checks/indicators.

In the specific case of AF447 there were other systems to indicate a problem, with the loss of airspeed indicators

The aircraft's stall warning sounded briefly twice due to the angle of attack tolerance being exceeded

Another indicator would have told the pilot that they were climbing too fast

By the time the pilot had control of the aircraft's roll, it was climbing at nearly 7,000 feet per minute

Furthermore, and much like any similar incident, it led to safer aircraft

On 12 August 2009, Airbus issued three Mandatory Service Bulletins, requiring that all A330 and A340 aircraft be fitted with two Goodrich 0851HL pitot tubes and one Thales model C16195BA pitot (or alternatively three of the Goodrich pitot tubes); Thales model C16195AA pitot tubes were no longer to be used.

The fact is, and it's one that is hard to grasp if you've not experienced it first hand, that spacial disorientation is really hard to recognize while in it. Pilots train hard to avoid it, but being humans they make mistakes. No amount of extra telemetry will ever change that.


No, a wind sock would not be effective as a replacement for nor as a backup to the airspeed indicator. What works well at wind speeds we're familiar with in daily life, say 10-30mph, wouldn't be the same at 500 mph.


You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .