It is probably a derived constant factoring in 2 things
1) density reduction due to consumption of sugar (reducing the density of the water)
2) density reduction due to production of alcohol (less dense than water)
there is a stoicheometric (i.e. constant) relationship between the consumption of one and production of the other, but after that it gets ugly... There are probably a lot of nasty differential equations relating to 3 phase systems (sugar, water, alcohol) and density, with ugly non-ideal-fluid approximations and virial coefficient power series expansions.
Just thinking about it is giving my brain thermodynamics cramps... time for beer
Be very thankful someone figured it out for us!