Unlike some people, the normal 120Hz flicker of fluorescent lamps is not something I notice (unless I look very closely at the ends) or bothers me but it bothers some people. What I do notice, is when a lamp is rectifying, swirling or blinking near the end of life of the cathode emitter. However, there are a few other "flickering" effects I don't quite understand, often related to cold temperatures and usually go away after the lamp runs a while:

-Striations: I see this often when a tube is operated on a HF and the tube is somewhat cold (even on CFLs). However, the same tube at the same temperature on a different ballast might not striate. Also, krypton (i.e. 34w/60w "energy saver") tubes also show striations on magnetic ballasts too, along with a slight pulsing of brightness at the same time. The only time I see it with argon tubes on magnetic ballasts is during the glow discharge phase (during rapid-start startup, on preheat with starter removed, or when a second tube on a series pair has failed), or under extremely cold (below-freezing) conditions for most any tube on a magnetic ballast.

There seems to be a bulb wall temperature vs. arc voltage relationship that leads to these striations.

-Pulsing: A slightly irregular dimming & brightening of most of the tube's length in cooler temperatures, seen only on magnetic preheat or rapid-start ballasts. I see this on F32T8's and PL lamps to a severe degree in very cold temperatures, but also F15T8's, F20T12's, etc to a lesser degree. Interestingly, 75w 8-foot slimline tubes pulsate at a very rhymthic rate, and rather noticeably, when cold.

-Shimmering: It looks a lot like rectification, but I notice it only on argon (i.e. full 40w) tubes on magnetic ballasts in cooler-than-normal-room-temperature conditions. I most often notice this with 40w tubes on rapid-start or preheat-gear.

Does anyone have an explanation for these discharge artifacts? Also, why does the same tube run perfectly when cold on an electronic ballast but pulsate or shimmer on a magnetic one?

When a 36W or 18W T8 switched on, I have seen two types of effects, which are analogous to each other in magnetic and electronic ballasts when the lamp warms up:
1. Standing waves (Striations): Occurs in HF ballasts.
2. Brightness vibrations (Pulsing): Occurs in magnetic ballasts.
Both effects occurs when the lamp warms up, located in a cold environment, or going mercury starved, depending to the ballast used.
F32T8 have the same argon gas as the F40T12. This is the 18W and 36W T8 that have krypton gas.

@arcblue:
The fluorescent tubes are very sensitive to the actual mercury pressure inside, what is a result of the cold-spot and/or regulating amalgam capsule (if present) temperature, so therefore it is highly sensitive to the operating temperature.

Striations happen, when some medium-high frequency (few to 10's of kHz) current happen to be in resonance with the tube as an acoustic resonator. The medium-high frequency current may be the main feed from the ballast (HF), or it is a result of the lamp-ballast starting to electrically oscillate by itself.

The pulsating effect may come from the starter taking some current from the tube during the reignition spike (when the lamp have to reignite after the current crossed zero; the lamp usually need higher voltage for that, mainly when cooler). The starter current is not sufficient for the starter to heat up and close, but it influence the time, when the lamp reignite, so some halfwaves the lamp light longer, some for shorter time, hence the pulsating. If this is the reason, removing the starter when the lamp is lit does stop this pulsating.

Or it may hapen, than each time the lamp is reignited, another point on the cathode become the arc root, so the arc would be "jumping" between two (or more) spots on the filament and this movement would be then perceived as a flicker.

On HF ballast there is no starter and the current change it's polarity so frequently, the ionization level does not follow, there is rather permanent arc with no reignitions at all. And with HF ballasts are starters usually not used...

The mains frequency ballasts are more prone to rectification, as after current zero cross the filaments swap roles. As the anode current may flow to different place of the filament than was the cathode arc root, the hot-spot may cool down, making the reignition later, so reduce the current in this direction (shorter phase angle). And as each of the filaments may behave differently in this matter, it may lead to rectification. Again, on HF the polarity changes so often, the temperature can not follow and the ionization persist over the current zero cross, these effects do not appear. Moreover most of the HF ballasts have a capacitor in series with the lamp, what allow the DC voltage to shift so it compensate for the cathode difference, so the current stay symmetrical.

@dor: Striations is not limited to the HF ballast, it does occur on the low frequency magnetic as well, the reason could be the lamp self-oscillating.

Here's one that I've seen a lot at work.  I think it's more of a ballast issue.  The light will flash on and off at a rate of about 1Hz.  I've seen several lights at work do it, and usually the only time I see them stop is when the maintenance men replace the ballasts in them.

Here's another interesting one: Sylvania "Cool White Plus" F40T12s (which start dim in the middle even at room temp.) flickering on a full power (.8 amp) rapid start ballast till they warm up...that same rectifying-ish flicker...

Is that somewhere or you can experiment with it ? If yes, is it with the ballast on many lamps, or with te lamps on many ballasts, or in combination of specific lamps and ballast ?

What happens if you flip the lamps ?

If you connect  end of each lamp (so that the filament preheat but lamps cant strike) is there any difference in behavio of the 2 filaments ?

i have 1 f96 2 tube slimline at the shop that gets a slight but very annoying strobing effect.lasts maybe 30 seconds at a time.
i first thought it was a bad lamp(darkened ends) but after it let go with a loud rumbling eol show with a very long hisssss and i replaced both lamps its doing it again.but i am using up  a pile of used lamps that were in the building when i got it.
i can handle replacing a bulb here and there but the strobing has to go.
i swapped tubes with another fixture and the strobing stayed with the fixture.
would be interesting to know what is causing this.
my easy way out is a motorola t8 ballast.

Can you trace the fixture for bad connections / see if it is related to the supply voltage dropping ? (tap a meter into the fixture inputs and compare when this happens vs nomal work)

Also see if tapping or hitting the ballast helps start o stop this, in this case it may be a bit of lose metal that gets near or farther from the core and changing the ballast impedance to the extent that it either goes close to saturation, or is underpowering the lamps due to increased impedance

I see this same-looking effect in underpowered 240v Switch Start circuits, the lamp tend to have this slight 50hz flicker when the voltage is below some margin. If there are few lamps in a room (and here each lamp is on a separate ballast, so 2xF36T8 fixture have 2 36w chokes), the flickering can be in sync in multiple lamps

already done all that.
i suspect the cap is going in the ballast.i autopsied one that was kinda dim and flickered.
cap was shorted.

@icefoglights : I've seen the 1Hz flicker as well on (Advance, I think) electronic ballasts at church. The only temporary solution was to remove the tubes, and the permanent solution was to change the ballast.

Very often some bad connection cause such flicker, mainly around thermally exposed parts: as the thing heat up, it loose contact, as it cool down, itreestablish it,...

The capacitor short may be of similar nature: The damaged material change it's conductivity with temperature, while it causethe power oscillator to stop. Then the voltage disappear, all the thing cool down and the oscillator restart again.

Generally anything, what cause the oscillations to stop would lead to the same, as the "starting" circuit have usually few 100's ms delay...