without doing a lot of digging ,its based on Eulers  buckling theorem for struts under direct load downwards.
I = Fx lo sqrd/(pi sqrd x E)where E is modulus of elasticity , lo =2x length of the strut, F is load slenderness ratio y is Sqrt 2xE/compressive stress allowed and is limited to 122 for steel
buckling stress = pisqrd xExI/(losqrd x Area) and must be less than F/A

 . the load would need to be uniformly located about the c/l . and the above only hold for free columns ie no restraints at the top

To answer the first part of your question, I believe the reason that there is the l/r limitation, is that for very large l over r ratios a small inadvertant eccentricity can make a large difference.

The slenderness "limit" of 200 is simply a recommendation meant to ensure a practical design with adequate stiffness for erection.  Very lightly loaded members can exceed the slenderness limit if the calculations pencil out.  However, I would try to stay within the recommended limit if at all possible.

Another aspect is that of testing.  Frank Lloyd Wright designed the Johnson & Johnson Headquarters building with very thin, tapered columns that flared out at the top.  The local building department of the city denied him a permit to build such columns stating that they were too thin.

Wright had the contractor build a test column to scale and then loaded it up with a huge number of sandbags.  There's a famous picture of it "out there" somewhere with a bunch of workmen sitting on the sand bags.  

The building department relented and allowed the building to be constructed.  If you have the time and money, you can design the thin column and verify its capacity by load testing.....but usually this takes up too much time and money.

when slender over 200, the design stress should less than 0.5*allow stress.

Thanks ALL.
   In my case I have series of T-shaped column 21 feet high with very light loads say 1 kips on top with small eccentric. With slenderness kept within 200 it comes up with too heavy section for me comparing to others such as street lighting.
   Burtoni your value based on experience, test or codes?

JAE:

It was the Johnson Wax Headquarters, but that is close enough!

I believe this is what you are talking about:

Located in Rachine, WI

rhodie:  Yes!  exactly!   sorry about getting drugs and wax mixed up.

That should be Racine, WI, (no "h", it must have been a finger slip) which is just south of Milwaukee, WI on the beautiful Lake Michigan shoreline!

Great picture.  Bye the way, while this building is still considered a design classic, the roof leaked from DAY 1.  Wright may have been a great architect, but if the detail didn't have anything to do with aesthetics, he wasn't too concerned.  He was also, from what I'm told by people who actually worked with him (salesmen, craftsmen contractors, etc.,) a pompus SOB!

Public TV had a special on a few weeks ago about another Wright classic design, "Falling Water".  The house that he designed was literally falling into the water of the stream it is built over.  So, there was a multi-million dollar fix to it's structural system.  

Yes, I remember reading an article about Falling Water.  It seems Wright complained to the structural engineer that the cantilevers had too much reinforcing steel in them.  Apparently, Wright was worried that all that rebar in the concrete would add too much weight to the cantilever!  

He had a wonderful design ability.....a whole "style" of architecture developed around his work, but as an engineer, the above example shows how little he knew about structural design and behavior.

I've heard it stated that Wright's genius was in that his simple geometric shapes could create complex (beautiful!) geometry in combination.  One can always identify the "masters" from the slew of "imitators and wanna-be's" that follow.  FLW was no exception to this pattern.  I worked for a commerical design architect, and most of his projects emulated Wright's style to some degree.  

The only reason I even knew about the columns in Racine (sorry about the earlier lazy-finger treatment) was that the aforementioned architect asked me to help him design a similar column concept for a building he was planning for.

We eventually did achieve the "look" of the columns, but we cheated by simply EFIS'ing bare steel columns with a facade.