The chain, not the sum
Twelve short tons at 160 gallons a ton is 1,920 gallons, or 7,268 litres off the press. Then five cuts land on it in order: 6 percent gross lees, 3 percent fine lees and ageing, 2 percent fining and cold stabilisation, 1.5 percent filtration, 1 percent on the bottling line.
Added up that is 13.5 percent. Run in sequence it is 12.86 percent, because the 1 percent at the filler is 1 percent of the 6,397 litres that reach the filler, not 1 percent of the 7,268 that came off the press. The gap is 0.64 percent of the press volume — 46 litres, 62 bottles, five cases. On a lot this size that is a rounding argument. On a 200 ton crush it is the difference between ordering the right number of pallets and being 90 cases short.
Losses do not run backwards
The mistake that costs real money is treating the chain as reversible. If a filtration pass takes 1.5 percent, adding 1.5 percent back to the filtered volume does not recover the pre-filter number: 1.5 percent of the smaller figure is a smaller quantity than 1.5 percent of the bigger one. To go from a case target back to a fruit weight you divide by the keep factor — here 0.8714 — and the page does exactly that, then runs the result forward again and prints what it lands on, so the round trip is visible rather than asserted.
Ask for 500 cases at 12 by 750 mL and the arithmetic wants 4,500 litres at the filler, 5,164 off the press, and 8.53 short tons of fruit. Run 8.53 tons forward through the same five stages and you get 500 cases back. If those two numbers ever disagree, one of them has been computed with an addition where it needed a division.
Where the yield figure comes from
Everything above hangs off gallons per ton, and that is the number this page most wants you to replace. It is not a property of grapes. It is a property of your press, your programme and your decision about how much of the press fraction you are prepared to put in the same tank as the free run. Whole-cluster pressed whites, destemmed reds pressed after a fortnight on skins and a bag press worked gently for a sparkling base are three different numbers on the same fruit.
Measure it once properly and you own it: weigh the bins in, and measure the tank rather than trusting the tank chart, which is often a nominal capacity rather than a calibration. The rest of the page is arithmetic and will be right the moment that figure is.
What it cannot see
Blending. Topping from one lot into another moves wine between lots without losing any, and this page has no idea that happened. Nor does it know about a lot declassified after tasting, fruit rejected on the pad, or a barrel written off. It also assumes every loss percentage you type in is of volume rather than of value, which is a fair assumption for lees and a poor one for the fraction of a run rejected at the final check, where you have paid for everything and recovered nothing.
The last thing worth saying plainly: none of this speaks to what you are allowed to do. Water, acid, sugar and every other addition are regulated separately in every country, labelling is regulated on top of that, and excise is a third regime again. The arithmetic here is indifferent to all three.
Questions people ask
How many cases of wine come from a ton of grapes?
At 160 gallons a ton and the five losses in the form, one short ton finishes at 58.6 cases of twelve 750 mL bottles — against the 67.3 cases the press volume alone would suggest. The whole difference is the cellar. The press yield is the dominant term and it is yours to measure: whole-cluster, destemmed and hard-pressed lots on the same fruit will not give the same figure.
Why can I not just add my loss percentages together?
Because each stage takes its share of the volume that reaches it, not of the volume you started with. Five losses of 6, 3, 2, 1.5 and 1 percent add to 13.5 percent but chain to 12.85 percent, since the filter is only ever working on what survived the rackings. The addition method double-counts, and the error grows with the number of stages and with how big they are.
How much fruit do I need for 500 cases?
Work backwards through the keep factor rather than adding the losses back. 500 cases is 4,500 litres at the filler; divided by the 0.8715 keep factor that is 5,164 litres off the press; at 160 gallons a ton that is 8.53 short tons. The page prints the forward run of that same weight underneath so you can see the two agree.
Which loss is worth attacking first?
The page names the biggest one in litres rather than in percentage points, which is not always the largest percentage — a small percentage early in the chain works on a bigger volume than a large one at the end. With the defaults it is the gross lees racking at 436 litres. Whether any of it is recoverable is a separate question about lees settling time, filtration of the lees, and what it does to the wine, and none of that is arithmetic.
Does this tell me what I am allowed to add or how to label the wine?
No, and it cannot. Additions, water, acidification, labelling and excise are set by national and regional law that differs by country and changes. Everything on this page is volume arithmetic on figures you supply. The regulator you answer to and a qualified enologist are the sources for what is permitted.