Just curious, A F20T12 fluorescent lamp has a rated average life of 9,000 hours. A full power F40(T12)has a rated average life of 20,000 hours. Therefore, a 40W lamp theoretically lasts 11,000 hours longer than a 20W lamp. Is this due to length? The 40W lamp has more emitter? Operating conditions e. g. preheat vs rapid start? What is everyone's thought?

To fully answer this, you need the way the mfgs. measure lamp life.

Same. From start untill half the lamps EOL

I think the main difference is the manufacture of the lamp itself - they want or dont want it to last. In the EU T12 is banned, so the manufacturers make the few remaining "special application" T12's rated just 6000..9000 hours, i guess in part to not make them attractive to the users except where there is no choice. In T8 the Halophosphor tubes are "undesirable" to them so they make them rated 13K hours while the Triphosphor ("normal" for EU) tubes are 20K

The fluorescent life is to a big extend related to the way, how it is used and mainly how frequently and how exactly it is started. With many ballasts the lamp life is actually limited by the number of starts and not that much by the hours burned.

With the F40T12 the life rating may be designed for their commercial use (as that was their major application), where a 12hour/start is quite a common pattern for life rating. With a 20khour life that means about 1800 of starts.
On the other hand the F20T12 is more often used on a simple preheat ballast at homes, where the 3huor/start pattern became the standard for lamp life rating. With the 9000hours rating it means about 3000 starts, so actually more than the F40T12 rated for commercial use.

@Ash:
That is, what conspiracy believers say.
The thing is, any technology, after became assumed as obsolete, never earned any funding for improvements or so.

The common T12 rating was in the range of 6k to 10khours. Then the more efficient T8 came and because they are possible to directly use in majority of installations, the T12 were meant to just fill the niche, where the T8's were not possible.
But all future development (improving the cathode life,...) was applied only for the newer T8's, so the 13000hour life rating was available only with these.
The same applied with the future development of the tri-phosphors: When the life was further improved, that improvement was implemented only for the new tri-phosphor types, so the 24khour rating came only with those.
Don't forget the cathode is quite extremely sensitive for the lamp atmosphere (the T12 used just Argon, which is by the way quite aggressive "sputtering agent"; the T8 are Krypton based, which is way easier on the cathodes).
And the phosphor mix and it's technology has great impact on the residual concentration of the contaminants like water (metal migration catalyst) or so.
So all lamps made just for the legacy installations were just made using the original design, so offering just the rating as was common at the time they were at the time of their "top glory". And that is the 9000 hours for T12, 13khours for halophosphate and it is up to the 80+khours for the tri-phosphor T8's.
Even when some of the modification would be possible to transfer back to the older formats, it would still mean quite some investment into development work (someone would have to figure out, if the given modification will ever work at all). With a niche product, with it's use place just disappearing from the world that will never pay off.

And maybe some manufacturers choose to rate lamp life in a scam kind of way, for X Hours with 5% of tubes surviving.

Interesting thing about the 13K hours. We ahd those Osram Basic 36W/765 here from the 80s. Up to mid..late 2000s they were made in Germany, and lasted 10+ years in everyday use. Now they are made in China, same lamp designation, same rated life, they last 3 years

In the US/UK there were T12 lamps rated >20K hours allready in the 70s. And last they did. In same places of use, a 90s T12 lamp (i am not that old to see the 70s ones, or significant quantities of 80s ones) lasted well above 10 years (school classrooms built 10 years ago, ~12 out of 14 lamps are original), same as the Germany T8. Modern T8s EOL every few years

10 years * 8 effective months/year (not counting summer vacation, holidays and weekends) * 30 days * 8 hours/day = about 20K hours

You can not judge anything related to longevity from just few pieces. So few tubes surviving 10+ years does not mean that technology could be 10year rated. You have to count all lamps, that were replaced already long time ago (and nobody remembers that anymore).
And other aspect is, what is the mortality rate at the given life rating and that is related to a mortality curve.
In the past quite flat mortality curve was quite common with many products, so even when they were rated just 9000hours, there was quite a lot of pieces surviving 20khours and longer. Yes, people tend to see these as "representatives", but they are not.

What have happened with modern manufacturing was mainly the mortality curve becoming rectangular. Why that? The mortality is a conjunction of many failure modes. Some of them are intrinsic to the technology (something gets consumed), in ideal world those are expected to cause EOL at exactly the same time for all pieces (any deviation means there was something wrong - like the "Livermore Century Bulb" was in fact defective since new, because had way lower efficacy than it was supposed to have). And then there are latent manufacturing defects, which tend to happen at quite a steady rate, in an ideal world they would never happen at all.
And it was the second type of defects, what became quite eliminated from the production. So before that, the latent defects caused half of the lamps to die before the 9000 hour mark (typical were mechanically broken cathodes on T12), but still some "lucky" pieces survived the 20khour.
With the modern T8 the defects were practically eliminated, so most lamps survive till the 20khours and then most fail within short time (because that is, when the emitter gets consumed).

The thing is, with many intrinsic life limiting effects the long life costs something else (e.g. with incandescent intrinsic long life means low efficacy), so a compromise has to be made. When the elimination of manufacturing defects significantly prolonged the median life, there was room to improve some other parameter, even when it means the life gets shortened a bit back. Very frequently just the change in the process responsible for elimination of the defects means the intrinsic life gets reduced a bit as well (too much of cathode mix dosing causes filaments to more likely to break prematurely, while still the amount of the cathode mix is the principal life limiter)

Whole school built in the early 90's, how is that for statistics ? 10 years later, the surviving lamps were ~12 out of 14 on average between classes. Easy to tell, because originals were T12 and replacements all T8. In another school, same statistic, originals were T8 Osrams, replacements various other T8's and brighter than the worn out old tubes, so still easy to see which tubes are which

Are you sure they were never replaced just by the T12 at that time?

In the school with the T12s, the replacements were always T8. The T12s were supplied by the contractor who built the building, T8s bought by the school

In the school with the T8s, they used few types of lamps for replacements, but none of them identical to the originals. Also, as the old ones kept dropping in light output when approaching 10 years, every newer lamp was really standing out by its light output