Ok my interpretation of the wiki is this: each planet is assigned a mean or average size, then the standard deviation is assigned. Then a normal distribution, or gaussian, or "bell curve" type random number generator is applied to the particular planet slot characteristics in such a way that 32% of the curves probable outcome falls outside 1 standard deviation either direction from the mean. This bell curve is adjusted with respect to the standard deviation in such a way that there is always a 32% chance that the outcome will be outside the "one standard deviation in each direction" regardless of the standard deviation. In other words there is always a 68% chance the planet will come in between the upper and lower limits of each planet slot as stated in the chart in the wiki, with the majority of these concentrated around the mean.

IF ,however, the gaussian number turns out to be one of these "32% that are outside one standard deviation in each direction", then a uniformly distributed, or purely random, or "flat curve" between 40 and 320 is used instead.

So you get a 32% chance to use the method that is "capable" of producing more than the upper limit for each planet, but then that method only yields a 42.8% of actually producing a planet larger than 200 fields. That translates to about a 13.6% chance of getting a planet larger than 200 fields, with an equal chance of getting a planet smaller than 160 fields.

It is similar to this. You get to flip a coin to determine if you win a penny or a million dollars. First you flip a penny, if its heads you win a penny if its tails you have to flip a quarter instead. When you flip the quarter if its heads you win a million dollars, if its tails you win a penny.