So you'd round one instance of 49.512% up to 51%, one instance of
49.512% down to 49% and 0.978% to 1%?

I can't make that work...

0.978% rounds up to 1%, leaving 99%. Since there are only two values,
one instance of 49.512% will have to round up to 50%, one down to 49%.
How should XL decide which one?

Or, if I use your logic with, say:

12.41%
12.41%
12.41%
12.41%
24.39%
24.39%
0.53%
0.53%
0.52%

0.53% is the highest fractional value, so it rounds up to 1%.
0.52% is next, so it rounds up to 1%
0.41% is next, so 12.41% goes to 13%

100% - 3*1% - 4*13% leaves 45%, so one instance of 24.39% will be
rounded down to 23% and one instance will be rounded down to 22%.

Is that what you're after?

In article ,
Curt J wrote:

My preferred answer is 51%,49%,0%.

My decision making criteria is that .903 & .585 are closer to 1 than
.512 so therefore I will round up on 50.585 and 48.903. Since .512
is smallest it will not get rounded up.

<snip

There isn't a universally right way to do this. For instance what about:

49.512%
49.512%
0.978%
-------
100.00%