Well, the frequency of a given length of a string is all related to the square root of the string tension divided by the mass per unit length. So for a thicker string, the mass per unit length is increased, so the tension also has to increase in proportion to give the same frequency. On wound strings, the core wire really provides most of the tension element, whilst the wrap provides most of the mass. Different manufacturers will use different combinations of core wire diameter and wrap thickness to give different tensions for the same overall string diameter.

The actual string length is always slightly hard to define exactly because (and especially with thicker strings), there is a small part of the string at the two fulcrum points that isn't really vibrating, so the overall string length becomes slightly shorter. This is one reason why the bridge saddles are normally further back than you'd expect from measuring the nut to the 12th fret distance and doubling it. As the non-vibrating length of string is always the same length for a given string, then the proportion of the string not vibrating is higher with the string fretted at the 12th fret, than with an open string - or conversely, the length of the vibrating string is proportionally shorter when fretted at the 12th fret, so the bridge saddle needs to be moved backwards to compensate.

It's an iterative process, because as you move the saddle back, the overall string length increases so you then need to increase the string tension, measure again, then adjust position, increase string tension, measure again etc. Almost impossible to get it perfectly in tune.

So maybe Waz, there's some slight variation in string thickness on your G string that means the bridge end of the string has a slightly higher mass per unit length than the nut end of the string. But without cutting it up into little equally sized pieces (a feat in itself) and weighing them on a very accurate scale, you'll never know for sure.