Chapter II · Grape chemistry and harvest
How much must one tonne of grapes yields, how it divides into free-run and press juice, how many grapes fit into a qvevri, and when the harvest should begin.
Yield is computed from berry mass (without stems). Conversion to litres uses must density computed from sugar (see "Sugars and must density"). Press juice is split into two fractions: first (2/3) and heavy pressing (1/3).
Mash volume: berries and pomace ≈ 1.09 kg/L, stems ≈ 0.8 kg/L; in the partial-pomace methods the must comes from a vertical press. A qvevri is not filled to the top — the pomace cap rises during fermentation. Final wine yield and heating are guides.
Sugar approaches its plateau exponentially: B(t) = B_max − (B_max − B₀)·e^(−rt); r is computed from the two measurements. Weather (rain, heat) changes the forecast considerably — repeat the measurement twice a week.
The harvest is the most intense weeks of a winemaker's year. Several practical questions must be answered: when to start picking, how many people and how much time we will need, how much must we will get and whether everything will fit into our qvevri. Mistakes here are costly: grapes picked too early give a sour, green wine, grapes picked too late a heavy wine that has lost its acidity, and surplus grapes are either sold or wasted.
Most of these questions rest on simple physics: mass and volume balance, density, the thermochemistry of fermentation and the kinetics of ripening. On this page we solve all three problems quantitatively: must yield, qvevri filling and the harvest calendar.
The oldest way of pressing grapes is treading by foot. In Georgia this was done in a satsnakheli — a long trough hollowed from a log or carved in stone, from which the must flowed directly into the qvevri. Stone satsnakheli and cellars with qvevri survive at Vardzia (12th century) and many other monuments. Foot treading is gentle: seeds and stems are hardly damaged, so the must is less bitter.
Beam and screw presses appeared in antiquity around the Mediterranean, and the vertical (basket) press was the mainstay of European winemaking for centuries. In 1951 the German company Willmes built the first pneumatic press: a rubber bladder inside a cylinder inflates with air and presses the grapes gently against the wall. Low pressure (0.2–2 bar) gives a high yield with minimal phenolic extraction, and it is now the standard for white wine.
For centuries the timing of the harvest relied on taste and experience. From the 19th century the hydrometer and refractometer were added, and in the 20th century analysis of acidity and phenolic maturity. Today the decision balances several criteria, and practical constraints — people, equipment, weather — are often decisive.
Stems account for 2–7% of the cluster mass; the rest is berries. Of the berry mass, 75–90% is flesh (juice and its cells), 5–12% skins and 2–6% seeds. In theory 80–85% juice can be obtained from the berry, but in practice part of it stays in the pomace — the less, the harder the pressing.
Free-run juice flows out as soon as the grapes are crushed, without pressing: usually 50–60% of the berry mass. The rest is press juice, released as pressure increases. The calculator computes must volume as:
$$V_{\text{must}} = \frac{m_{\text{berries}}\cdot \eta}{\rho}, \qquad m_{\text{berries}} = m\,(1 - s)$$
where $m$ is the grape mass, $s$ the stem share, $\eta$ the total juice share of berry mass (0.70–0.85 depending on the press), and $\rho$ the must density computed from sugar: $\rho = SG \cdot 0.9982$, $SG = 1 + B/(258.6 - 0.8796B)$ (see "Sugars and must density"). The pomace mass (skins, seeds and retained juice) is $m_{\text{berries}}(1 - \eta)$.
The fractions differ in composition. Free-run juice is the "cleanest": its acidity is high, pH low and phenolics few. As pressure rises, the skin cells break and more potassium, phenolics and colloids pass into the must, so press juice has a pH 0.1–0.3 higher, lower acidity and a coarser, more bitter taste. The heavy-press fraction may also carry green tones from stems and seeds.
In quality winemaking the fractions are therefore collected and fermented separately: free-run and first press go into the main wine, while heavy press goes into a simpler wine or is partly blended back after suitable treatment. In the Kakhetian method the question matters less: the whole mash ferments together with the pomace, and extraction happens by maceration rather than by pressing.
The volume of mash per kilogram of grapes in a qvevri depends on the method. In the Kakhetian method the whole crushed mass, stems included, goes into the qvevri:
$$v = \frac{1 - s}{\rho_{\text{berry}}} + \frac{s}{\rho_{\text{stem}}} \approx 0.93\ \text{L/kg}$$
($\rho_{\text{berry}} \approx 1.09$ kg/L, $\rho_{\text{stem}} \approx 0.8$ kg/L). In the partial-pomace methods the must is pressed first and a share $f$ of the pomace is added back: the Kartlian method (f ≈ 1/3, with stems) gives about 0.76 L/kg, the Imeretian (f ≈ 1/10, no stems) 0.69 L/kg, and the European (must only) 0.67 L/kg.
A qvevri is not filled to the top: during fermentation CO₂ bubbles carry the pomace to the surface, a pomace cap forms and the mash grows by 10–20%. A qvevri is usually filled to 80–85% of its volume; after fermentation, when the pomace has settled, it is topped up with wine and sealed. The grape mass that fits in a qvevri: $m = V_{\text{qvevri}}\cdot \phi / v$, where $\phi$ is the fill level.
Alcoholic fermentation is exothermic: converting one mole of glucose into ethanol and CO₂ releases about 98 kJ of heat, i.e. 0.546 kJ per gram of sugar. If the heat went nowhere (adiabatic conditions), the must would warm by:
$$\Delta T = \frac{S \cdot 0.546}{\rho \cdot c_p}$$
where $S$ is sugar (g/L) and $c_p \approx 3.85$ kJ/(kg·K) the specific heat of must. For 220 g/L of sugar $\Delta T \approx 28$ °C — from 15 °C to 43 °C, which would kill the yeast. In reality heat is lost: the walls of a buried qvevri and the surrounding earth absorb it, and the CO₂ also carries away water vapour. So the temperature in a qvevri usually does not exceed 28–32 °C. The larger the qvevri, the smaller its surface-to-volume ratio and the more strongly the mash heats up.
One gram of sugar releases 0.489 g of CO₂. One litre of must at 220 g/L of sugar gives about 108 g of CO₂, i.e. 59 litres of gas at 20 °C. A 2000-litre qvevri releases more than 100 m³ of CO₂ during fermentation. CO₂ is heavier than air, collects on the cellar floor, and concentrations above 4% are life-threatening. The marani must therefore be well ventilated during fermentation, and the air checked before anyone climbs into a qvevri.
After véraison sugar accumulates rapidly in the berry, but near ripeness the process slows: import from the phloem declines and Brix approaches a plateau (usually 24–27) asymptotically. An exponential approach describes this well:
$$B(t) = B_{\max} - (B_{\max} - B_0)\,e^{-rt}$$
From two measurements the rate constant is $r = \ln\dfrac{B_{\max} - B_1}{B_{\max} - B_2}\big/\Delta t$. The target $B^*$ is reached $t^* = \ln\dfrac{B_{\max} - B_2}{B_{\max} - B^*}\big/ r$ days after the second measurement. Linear extrapolation always underestimates this time — the forecast date comes out several days too early.
Harvest duration: $n = \lceil m / (N \cdot q) \rceil$ days, where $N$ is the number of pickers and $q$ the daily quota per picker (usually 300–800 kg by hand). The calculator places the harvest so that its middle day falls on the target date: some grapes are picked slightly early, some slightly late.
| Quantity | Typical value | Note |
|---|---|---|
| Stems in cluster mass | 2 – 7% | depends on variety |
| Skins in berry mass | 5 – 12% | more in thick-skinned varieties |
| Seeds in berry mass | 2 – 6% | 1–4 seeds per berry |
| Free-run from berry mass | 50 – 60% | depends on crushing |
| Must yield | 60 – 75 L / 100 kg | by press type |
| Wine yield (Kakhetian) | ≈ 55 – 60 L / 100 kg | after racking |
| Qvevri fill during fermentation | 80 – 85% | room for the pomace cap |
| Fermentation heat | 0.546 kJ/g sugar | ≈ 98 kJ/mol glucose |
| CO₂ | 0.489 g/g sugar | ≈ 59 L of gas per 1 L of must (220 g/L) |
| Sugar accumulation near ripeness | 0.1 – 0.4 Brix/day | slows toward the end |
| Equipment | Juice share of berry | L / 100 kg (22 Brix) | Character |
|---|---|---|---|
| Satsnakheli (foot) | ≈ 0.70 | ≈ 62 | gentlest, fewest phenolics |
| Vertical (basket) | ≈ 0.76 | ≈ 67 | traditional, slow |
| Pneumatic | ≈ 0.80 | ≈ 70 | gentle, high yield |
| Continuous screw | ≈ 0.85 | ≈ 75 | fast, harsh, more phenolics |
"Sugars and must density" is the basis of this page: density converts mass into volume, and sugar determines fermentation heat and CO₂. "Acids, pH and the buffer system" explains why press juice has a higher pH (potassium), and "Phenolic compounds" why the pomace share in the qvevri is the main lever of wine style. "Ripening dynamics" in Chapter I describes sugar accumulation through temperature sums; here the same process is forecast from field measurements.
Chapter III (Alcoholic fermentation) covers heat exchange during fermentation in detail, and Chapter V (The qvevri method) the preparation and filling of qvevri and pomace management.
The Georgian rtveli is traditionally a family and community affair: relatives and neighbours pick together, tread the grapes in the satsnakheli and pour the must into the qvevri the same day. Harvest planning is therefore often tied to the number of people and to weekends rather than to exact sugar. The calculator helps reconcile the two: if the target date falls on a working day, you can see how many Brix lower or higher the grapes will be a few days earlier or later.
Qvevri in a marani often differ in volume, and some are very large (2–3 tonnes). In a large qvevri fermentation runs hotter and the pomace cap is thicker, so it is sometimes filled less or the pomace is punched down more often. With the calculator you can work out in advance how many grapes fit in each qvevri and whether any will be left over.
| Term | Definition |
|---|---|
| Rtveli (harvest) | Picking and processing the grapes |
| Crushing / pressing | Opening the berries and releasing the juice |
| Free-run juice | Juice that flows out without pressing |
| Press juice | Juice obtained under pressure, in fractions |
| Satsnakheli | Traditional trough for treading grapes by foot |
| Chacha (pomace) | The mass of skins, seeds and (sometimes) stems |
| Destemming | Separating the berries from the stems |
| Pomace cap | The layer of pomace that rises to the surface during fermentation |
| Fill level | The share of qvevri volume occupied by the mash |
| Adiabatic heating | The temperature rise if no heat is lost |
| Marani | The wine cellar where the qvevri are buried |
A winemaker has one tonne of Rkatsiteli at 22 Brix, with a stem share of 4%. With a pneumatic press (free-run 58%) the calculator shows 511 L of free-run juice, 129 L of first and 65 L of second press juice — 705 L in all, or 70.5 L per 100 kg. The pomace is 192 kg, the stems 40 kg, and the must density 1.090 kg/L. With a satsnakheli the same grapes would give about 62 L per 100 kg, with a screw press 75 L.
The next year the winemaker makes the wine by the Kakhetian method. The marani has four qvevri: 2000, 1500, 1200 and 800 litres. At 82% fill and a mash volume of 0.93 L/kg they hold 4846 kg of grapes; there are 5000 kg on hand, so 154 kg are surplus. About 2760 L of wine can be expected. Adiabatic heating is +28 °C, so temperature must be controlled at the peak of fermentation. About 197 m³ of CO₂ is released during fermentation — the marani needs ventilation.
To plan the harvest: on 25 August Brix was 18.0, on 4 September 20.2. With a sugar plateau of 26 Brix the calculator gives r = 0.032 day⁻¹, a current accumulation rate of 0.19 Brix/day, and the 22.5 Brix target is reached on 20 September — in 16 days. A linear forecast would have said 10 days, almost a week too early. Eight pickers (500 kg a day each) need three days to pick 12 tonnes: the harvest starts on 19 September and ends on the 21st.
1. Open the "Calculator", mode "Must yield". Enter the grape mass and sugar and choose the pressing equipment. The cards show how many litres of free-run and press juice you get and how much pomace remains.
2. The right-hand chart compares all four presses on the same grapes. If you need more must, a stronger press gives more — but press juice is coarser, so collect the fractions separately.
3. In the "Qvevri filling" mode choose the winemaking method and type your qvevri volumes separated by commas. The calculator tells you how many grapes fit, how many will be left over or missing, and how much wine you will get.
4. Pay attention to the CO₂ card. During fermentation hundreds of cubic metres of gas are released in the marani. It collects near the floor and is life-threatening. Keep doors and windows open during fermentation, and check the air before climbing into a qvevri.
5. In the "Harvest calendar" mode enter the dates and Brix of two measurements, the target sugar, the amount of grapes to pick and the number of pickers. The chart shows the forecast and the harvest window. Repeat the measurements twice a week.
1. The juice share for each press is an average. Real yield depends on variety (skin thickness, flesh firmness), ripeness, enzyme use and the press programme. To calibrate with your own data, adjust the "free-run share" and the press type until the result matches your measured yield.
2. In the qvevri calculation, the must in the partial-pomace methods is assumed to come from a vertical press. Pomace volume varies with its porosity and compaction, so in the first year start by filling the qvevri to 75–80% and watch how far the cap rises.
3. Adiabatic heating is an upper bound. The real maximum depends on the rate of heat loss, which is set by qvevri size, ground temperature and fermentation rate. In large qvevri (over 2 tonnes) measure the fermentation temperature daily and punch down the pomace more often if needed.
4. In the harvest forecast $B_{\max}$ is the most sensitive parameter. If you have data from previous years, use the actual maximum for this plot. In hot, dry years sugar keeps rising through berry shrivelling (concentration) and $B_{\max}$ increases; in rainy years Brix can temporarily fall through dilution.
5. Sugar should not be the only criterion for the harvest date. Assess acidity and pH (the "Acids, pH and the buffer system" page) and phenolic maturity (the "Phenolic compounds" page) together, especially for grapes destined for the Kakhetian method, where seed maturity is decisive.
1. Why is the pH of press juice usually higher than that of free-run juice?
Potassium sits mainly in the skins. As pressure rises the skin cells break and potassium passes into the must — part of the acids becomes salts and pH rises. That is why fractions are often fermented separately.
2. A 2000-litre qvevri is filled with must at 220 g/L sugar. Roughly how much CO₂ is released during fermentation?
1 g of sugar gives ≈ 0.49 g of CO₂: 220 g/L → 108 g/L of CO₂ ≈ 59 L of gas per litre of must. 2000 L → ≈ 118 m³. That is why ventilating the marani is a matter of life and death.
3. Why is a qvevri not filled to the top for fermentation?
Gas bubbles carry the pomace to the surface and the mash volume grows by 10–20%. Without free space the mash would overflow the qvevri.
4. Brix rose from 18.0 to 20.2 in 10 days. Why can't the date of the 22.5 target be extrapolated linearly?
Sugar import from the phloem declines at the end of ripening, and Brix approaches a plateau (usually 24–27) asymptotically. A linear forecast gives a date several days too early.