I always thought they're sign of dying tubes (very old ones) or possibly of those started for the first time but today I saw them at a place where the tubes are quite new while already "burned in".
(It's a 05/2015 opened metro/subway station - well they might be using them while finalizing the station interior for a few moths before.)
The tubes in question (closed in translucent luminaires, here's a visualization) look like 58W/T8, probably electronically ballasted (not sure) and might be Spanish GE Polyluxes (that's what Prague public transport are mostly using these days).

All the tubes at the platform were showing clearly visible strips (in about 5 cm / 2 inch spacing), moving about 50 cm / 20 inch per second in the same direction. I wonder what might cause that phenomenon? Could it be rails powering?

The tubes are striating. That normally occurs with 34 watt energy savers here.

Standing sound wave in the Mercury vapors inside - spots with different pressure light at different brightness. The wave may be not perfectly standing but nodes are appearing at 1 end and disappearing in the other, making the pattern "run" - though it is still pretty much standing compred to the speed of sound (and especially the speed of sound in near-vacuum, as is the case in the FL lamp, which is faster than in air)

That's interesting, Ash. Didn't know about that.
What do you think causes that "sound wave"? Is it an internal cause (i.e. electrode vibrations due to a flowing current - that's why electric heaters are humming) or is it coming from outside? In the case above, there wasn't an obvious external cause - only a slight hum of ventilation etc.
It's also interesting that "striating" normally occurs on almost EOL tubes (at least what I know).

I've observed it on many occasions. I've noticed it on two types of situations : a dimly glowing lamp that has reached EOL a North American rapid start ballast and a lamp that is operated in a low temperature environment. Regarding the low temperature, it happens to some lamps in open strips that are too close to a HVAC vent, and to energy saver lamps that have not warmed up sufficiently. If the room is cold and lit with open strips containing energy saver lamps (such as F34T12 on rapid start ballasts in N. America) the striations can last seemingly forever. I've never seen it on European 36W T8 lamps, though (regardless of ballast type). I suppose it could happen in a cold environment.

I've seen this on 36W T8 during run-up to full brightness. This looks different with magnetic and electronic ballasts: On magnetic this looks like brightness vibration, and on electronic this looks strips that moves slowly and sometimes don't move at all.
I've also seen this with EOL T12 on Eltam Perfekt-Start ballast or T12 and short T5 glowing dimly directly from the mains without a starter or lamps that dims out as a result of impurities or mercury starvation.

That is how i understand it at least....

The origin of the wave is probably magnetic - The "arc" can "bow", and it does from the forces one section of the arc inflict on the other, probably starting with some initial imperfection at the cathode

Then the length of the tube act as resonance chamber. The tube then can resonate to the wavelengths, where distance between 2 nodes (zero crossings) of the standing wave line up with the length of the tube :

X = 2L (1 half wave)
X = L (2 half waves)
X = 2L/3 (3 half waves)
..
..

The matching frequencies are f = nc/2L where n=1, 2, 3....infinity and c = speed of sound in the Krypton / Mercury vapor at the present pressure (faster than in air due to low pressure)

One of that frequencies line up with the ballast driving frequency, and resonance starts.... As the lamp is not a perfect chamber (complex shape of the end walls near the electrodes, so not perfectly defined length of the chamber), it will react to a close frequency too

The "running" effects may be related to the frequency of the ballast changing (for example in sync with 100 Hz ripple on the main capacitor, or as reaction to something in the lamp), so the lamp being in constant changing between the 2 closest resonance frequencies : Alternatingly 1 extra node appears (step to next highr frequency) and 1 node disappears (step to the lower frequency)

If in ths lamp the node that appears tends to appear at 1 end and the one that disappears tends to disappear at the other, there is the running effect...

But i seen also lamps where the running pattern change directions, speed up and down, or nodes converge from both ends to the center (with the point of disappearing node being a bit random each time but around the center) and such, or just stay still

I've never seen it on European 36W T8 lamps, though (regardless of ballast type). I suppose it could happen in a cold environment.

The ambient temperature might be about 25°C/77°F. The tubes were probably 58W (definitely longer than 36W) though.

This looks different with magnetic and electronic ballasts: On magnetic this looks like brightness vibration, and on electronic this looks strips that moves slowly and sometimes don't move at all.

I think the striating can occur also on the magnetic ballast provided the tube is almost EOL.
In this case, it was probably an electronic ballast because the station is brand new and EU Ecodesign directives wouldn't allow classic chokes.

@Ash: That's an interesting theory! The fact is I've never seen striating circlines. It seems that only linear tubes do so.

EU directuives allow the manufacture of chokes untill 2017. I guess that means that they also allow those chokes to be used in the construction of lanterns, and those lanterns to be installed....

So i dont see how it makes sense to ban the use of lanterns allready now

There is the discourage game (make it green, be green, etc) that is played everywhere, but i guess it have nothing to do with actual regulations.....



Besides, the regulations are not based as much on ballast type (though they do treat Electronic and Magnetic ballast types separately), they just use some formula to define what is "efficient" or "inefficient"

In short, in the 2017 plan a 36W lamp at full power, complete with gear is "allowed" to draw ~38.5W - so ~2.5W losses. Present day "standard" 36W chokes have ~6.5W losses, and old "mini size" chokes closer to 10W

But there is a surprise waiting for the EU..... https://www.vossloh-schwabe.com/uploads/tx_sbdownloader/VS_KVG-Starter-System_2012_GB.pdf

The newer directive specify only the losses, so the low loss choke types still fit into the A2 (the highest efficiency rank category still without the dimming) are still OK...
The only difference is, the ballast will become heavier...

What about lamps that, according to their beliefs, can only work with chokes ie PL-S 2pin ?

They dont believe that those lamps can work on HF, despite common sense and evidence proving wrong

Would they accept that the chokes can be made efficient, especially so for low current/high voltage lamps like PL-x, and especially in the low wattages where even the A2 requirement only requires 3/4 efficiency ?

What about lamps that, according to their beliefs, can only work with chokes ie PL-S 2pin ?

They dont believe that those lamps can work on HF, despite common sense and evidence proving wrong

Would they accept that the chokes can be made efficient, especially so for low current/high voltage lamps like PL-x, and especially in the low wattages where even the A2 requirement only requires 3/4 efficiency ?

Who do you mean by "They"?
The EU directive does not make any difference between electronic cs magnetic, the same minimum efficiency applies for both. So if the F36 magnetic has losses of 2W and some electronic 10W, the electronic would be the banned and magnetic the allowed one from these two (once other than present "A" become banned). It is true, it is way easier to meet the efficiency limit by the electronic, but the directive does not care about this.

What it distinguish is the ability to reduce power when full light is not required, so in other words the dimming.
To reach A2, the ballast needs to feature the ability to reduce it's power consumption by dimming. Again, it does not matter if it is HF electronic or series choke with a heater transformer, both could be A1 if they meet the required minimum efficiency level (by the way it is a bit relaxed compare to A2 if it features the dimming functionality) and the dimming reduces input power.

I am asking about the lamp bans, not ballast bans

I read around here that lamps that are "not capable to run on efficient ballasts" will be banned. That was said at an earlier stage when the EU was defining ballasts by type, ie implying that Eelectronic are to stay and Magentic are banned. The 2 pin PL-S is intended to work on Magnetic only, and as such, it would by that definition be banned

(Allthough in fact 1. The ballast may be made efficient, 2. The lamp can in fact be used on Electronic too)



Actually the efficiency levels are differently defined, as i understand :

A2 - Minimum required for non dimming ballast

A2 BAT - Higher grade for non dimming ballast

A1 BAT - Minimum required and only grade proposed for dimming ballast, at full power being as efficient as A2 (non BAT)

I understand it as there are "efficient" (that means A2) ballasts for these, so the lamps are OK from that perspective.
With the HF ballasts and PL-S: Most combinations work, because most makers do use practically the same capacitor values. But the standard for the PL-S is written so, it does not specify the capacitance at all. So if a cheepeese maker does not use any capacitor at all (and pretend that as still OK from the RFI and starter arc quenching perspective; and I've seen many lamps like that), such lamps won't work on the HF ballast, yet still meet the standard.
And because these lamps are assumed outdated by the makers, there is no incentive for them to agree on any standard update, as today they have full freedom in selecting the value...

I have not seen any Electronic ballasts that are rated for 2 pin PL-S, and no attempts to make an effcient Magnetic either, allthough it is definitely possible and would be even simpler than for larger lamps (as the PL-S are low current lamps, so ballast losses are lower to begin with)

How about not relying on the capacitor at all, but just provide HF with inductive impedance - So it starts with the glow starter same as on 50 Hz ?