The alcohol refractometer has arrived but didn't work as expected. It turns out it is only suitable for spirits. I tested 40% abv white rum and got an accurate reading. Corvoisier brandy (40%) read 43%, Tequila and Bells whisky (40%) 38%. Curiously, Grand Marnier and Cointreau (40%) didn't register at all, whereas ouzo (40%) registered 47%!
However a range of wines with known strength produced results in the range of 18-22.5! A pattern emerged. When you subtract 9 you get the right result. This is because wine contains many ingredients, which are removed by distillation, which influence the refractive index.
Even a dry wine may contain up to 2% unfermented sugar plus acids, glycerine and so forth. Fortunately I have a sugar refractometer which I use for testing grapes. For all the wines tested with this I got consistent readings of 7-8 brix. As the wines were dry/medium dry, clearly they did not contain 7-8% sugar, as the reading suggests.
In order to establish how much alcohol affects the specific gravity, I tested the rum and got a reading of 14 brix, roughly one third of the abv and the equivalent of sg 1055. Divided by 4, the results would be 10% abv and 3.5 brix, sg 1013. I have read that a grape sugar refractometer becomes progressively less accurate at measuring sugar below 15 brix due to other elements in the juice.
Simply subtracting 9 from the alcohol refractometer readings of wine is good enough for me but there are cases where this number should be lower or higher than that. For example fino (dry) sherry at 15% abv read 22% abv and brix 7, which gives the correct result of 15% abv. A dry dandelion wine made with a low grape content (20%) read 19% abv and brix 7, giving 12% abv, also correct. Other dry/medium dry wines with a low fruit content (which include most kit wines) gave a brix reading of 7, whereas true wines (100% grape) read 8, with no difference between reds and whites, except the fino sherry and sake, but the dry white wine required a deduction of only 6 to give the true result. This is because reds are heavier.
However, when I tested port (20% with a high sugar content) I got a reading of 60% abv and 18 brix, which gives 42% abv, more than double the true value! My 'Sauternes' (sweet) read 30% and 11 brix, giving an unlikely 19% abv, (although 18% is possible without fortification). Pure mead (medium, 12% abv) gave 23%abv and 9 brix, giving 14% due to the higher sugar content, so 11 needed to be deducted. On the other hand, Pimms (25% and dry) read 40% abv and brix 14%, which gives about the right result.
The lowest reading was my sake (pure rice wine - no grape or added sugar) was 17% abv and brix 7, giving 10% abv. This is usually brewed to 18% abv but sold at 15%. All I can say is that mine certainly tastes a lot stronger than 10%! The lowest known strength tested was medium dry perry at 7.5% abv, which read 18%abv and 7 brix. To get the correct result, 10.5 would need to be subtracted which suggests that as the actual abv falls, the greater the correction required, but then 7.5 is not strictly speaking wine anyway.
