Article on The Internet Mini Encyclopædia 
Very simplified and very exaggerated example of cornerweights
"Some of you will say that this is stupidly exaggerated, but it does make some points blatently clear in my mind.
1. Take a blatently ordinary table that has NO flex or give, and has infinitely accurate legs and it sitting on a completely flat surface. To start with, all the legs will be touching the ground with identical force.
2. First off, lets sit at one end of the table table and call it the engine end. Now, all 4 legs are still in contact with the ground, but there is more force on the front 2 than the rear 2.
3a. Now we will shave off a mere 0.1 MM from one of the legs at the other end of the table. Suddenly it is doing nothing, NO weight is on this leg BUT the table will not have changed its attitude, it will still be level because of you sitting on the other end. Interestingly, the others must also now be supporting more weight.
3b. Lets go back a step now so we are at the end of step 2 again. This time, we will remove 0.1mm from one of the legs at the end you are sitting at (front). This time, because of your weight, the table tips, there is still weight on the fronts, but the (diagonally) opposite one at the rear is off the ground and has no weight on it. This however DOES affect the attitude of the table. So thats all pretty simple, basic theoretically perfect table physics with no springs.
4. Now we have a similar table, but it has sprung legs. You sit on the front again, and the fronts are adjusted relative to the rears, so that it is level.
5. Now we will again shorten one of the rear legs. If you are heavy enough, the table will hardly move at all, but the leg will still droop down to the ground, but with less force. If you are less heavy, due to the force of the diagonally opposie leg, the table will probably tilt towards the leg you shortened. This could be corrected in 2 ways. You could a. put the rear leg back to where it was, or b. shorten the diagonally opposite leg until all is level again. Now imagine if you did "b." then neither of these 2 legs have much force on them.
It is quite possible to end up with such an arrangement if you just measured the distance from the table top to the ground on each corner. "Its level, so whats the problem"
6. Take this well balanced table that has just had its ride heights set with a tape measure and is spot on level. Put it on a set of scales and prepare to have a shock. The 2 diagonally opposite legs that you did not touch have, between them, well over half the weight of the table (incl sitter). The other 2 however have well under half of the weight between them!!! Even though the table is spot on level, the cornerweights are well out.
Now transfer this setup to a Car (a front wheel drive one for this example). You can get a very similar effect. The car can appear to be level, but 2 diagonal corner weights can be set way too low. The amount of adjustment needed to get these errors is minute! 1/2 a turn on a hilo can throw things WAY out!
If you drive a car like this, the diagonally opposite wheels get worked a lot more than the other two which is not a good thing for handling.
The ONLY way you can get it correct is by setting the cornerweights. The only accurate way of doing this is with a set of 4 scales.
There are a number of explanations of cornerweights going round, some ignore the diagonals, some swear by it. Diagonal sums is the only one that makes complete sense to me and is what I swear by (Obviously trying to get the general balance correct too).
I hope this makes it a little more clear.. I think it is an OK description, OK simplified, but the basic principle is the same and it DOES transfer to cars in a very similar way.
BTW, my theoretical table does not have shock absorbers, preloaded legs, anti roll bar etc ;)"
A practical example and some more theory
OK, basically it is easiest with 4 sets of scales capable of up to 250kg, but at a push you can do it with 2:
I will use some numbers to explain what happens...
Plan diagram of weights at wheels
FL 194  FR 233  
66%


83%


RL 161  RR 156 
Total


L=355  744  R=389 
From this we can see that the Rear Left has 83% of the
weight that the FL has, and the RR has 66% that the FR has.
The left side of the car = 355 and the right = 389.
Apparently the difference between left and right is not that critical, but it is
important to match front/rear % of each side.
Looking again there is a big difference between the fronts, the FR is pushing
down considerably harder than the FL, while the rears are close the RL is still
a bit heavier. We need to move the left weight forwards and the right weight
backwards to get them nearer.
If you imagine (as the explanation above suggests) that this is a table with
unsprung legs, then the highest weight is the longest leg. That would mean the
table is unstable diagonally because it will tend to rock about a line from FR
to RL. To correct this we and make the short legs longer  or the long legs
shorter.
The relationship is a little harder than with the table as 2/3rds the weight is
carried over the front wheels
A quick test of pushing down on 1 corner shows how it affects the others! If the
RL is pushed down, making it heavier, then the FR gets a little lighter, the FL
gets a little heavier and the RR stays almost the same.
As the rear of the car is a little low we decided that lowering the FR a bit (as
an arbitary value a couple of mm was used). With the FR supporting less weight
the FL should get heavier, the 'table' should also tip towards the FR making
it's diagonal  the RL lighter. Well that was my theory :)
Then drive the car aroung on the flat and measure again twice to reveal:
FL 202 (+8)  FR 215 (18)  
73% (+7)


78% (5)


RL 158 (3)  RR 158 (+2) 
Total


L=360 (+5)  733 (11)  R=373 (16) 
Well pleased with the results, the front/rear % of each side had balanced up nicely. Also the L/R weights got closer too, which can't be a bad thing.
Join the Minilist share your joys and woes with 500+ mini owners worldwide via Minilist at Yahoogroups 