Reading Gravity, ABV & Efficiency

Three numbers tell you most of what happened on brew day and in the fermenter: original gravity (OG), final gravity (FG), and the efficiency you got from your grain. From OG and FG you get ABV and attenuation; from OG and volume you get brewhouse efficiency. Measuring them is what turns brewing from guesswork into something you can repeat on purpose.

What specific gravity is

Specific gravity (SG) is the density of wort or beer relative to water (1.000). Sugar makes wort denser than water (say 1.050); as yeast eat the sugar and make alcohol — which is lighter than water — gravity falls toward the FG (say 1.010).

Brewers talk in gravity points: the digits after the decimal × 1000. So 1.050 = “50 points”. Points make the maths easy.

The two tools

Hydrometer

A weighted float — the more sugar, the higher it sits. Cheap, accurate, and works right through fermentation. Two things to get right:

• Temperature. It’s calibrated at one temperature (usually 20 °C); a hot sample reads low. Cool the sample, or correct (~+1 point per 6 °C above 20 °C).
• Read at the bottom of the meniscus, with a CO₂-free sample (bubbles cling to the float and read high).

Refractometer

Reads sugar from two drops as °Brix — perfect on brew day because it needs almost no sample and no cooling. Two catches:

• Wort correction. Most read slightly high on wort; divide Brix by ~1.04 (or use a calculator) before converting to SG.
• Alcohol breaks it. Once fermentation starts, alcohol bends light and the reading goes wrong. Refractometer for brew day; hydrometer for fermentation — or use a refractometer-FG calculator that corrects using OG + current Brix.

Tilt hydrometer (RAPT Pill, Tilt)

A sealed float that reports continuously over wireless, which is what makes a “flat for twelve hours” trigger practical at all. It solves gravity from the angle it floats at, and that is the source of every failure mode it has.

• Recalibrate every time you open it. Charging means unscrewing the body, and reassembling changes its internal balance. A shifted balance shifts every reading it will ever give.
• Calibrating on the bench does not reproduce floating. The device reads its full accelerometer vector, not pitch alone. Resting it at “about the right angle” on a towel fixes one axis and leaves roll and yaw wherever they fall. That is exactly how you get a number that looks plausible on the bench and is thirty points wrong in wort.
• The one-minute check: float it in plain water and expect 1.000. Do this whenever the device comes out. It is the only test that catches the problem before the readings matter.

Getting into the calibration screen (RAPT Pill)

The configuration portal only comes up while the Pill is plugged into power, but calibration needs the device floating freely and unplugged, because it has to read the angle it actually sits at. The two requirements look contradictory and are not:

1. Plug the Pill into power and let the portal open.
2. Navigate to the calibration screen before unplugging.
3. Unplug. The portal stays open for about ten minutes, which is the window to calibrate in.

If the portal drops as soon as you unplug it, charge the Pill to 100% and try again. A full charge appears to reset something in its power handling, and from then on it holds the connection through the unplug as it should. This is not a mid-brew fix, so do it before brew day rather than discovering it with wort in the kettle.

A constant offset can be corrected afterwards; an attitude error cannot. If the device is simply reading six points high all the way through, every figure can be recovered by subtraction. If the float attitude was wrong, the error distorts the span as well as the zero, so there is nothing to back out and the readings should be discarded rather than salvaged.

Two faults can cancel and look like agreement. Temperature compensation misbehaves well below the range these devices are calibrated over, so a unit sitting at 3 °C with a six-point positive offset can read almost exactly right, and appear to confirm a glass reading it agrees with for entirely the wrong reason. Two faults roughly cancelling is indistinguishable from no faults at all. Check the device in water rather than trusting the agreement.

Even a badly calibrated one is still useful. The absolute number is wrong, but the shape of the curve is not, and most decisions during fermentation are shape decisions: has it flattened, has it started falling again, has the rate of change rolled over. Let the tilt do the watching and let glass supply any number that goes on the record.

ABV

The simple, good-enough formula:

ABV (%) ≈ (OG − FG) × 131.25

Example: OG 1.052, FG 1.010 → 0.042 × 131.25 ≈ 5.5 % ABV.

For stronger beers a more accurate form exists, but at normal ale strengths the simple one is within a rounding error.

Attenuation — how dry it finished

Apparent attenuation (%) = (OG − FG) / (OG − 1.000) × 100

Example: 1.052 → 1.010 gives (0.042 / 0.052) ≈ 81 %. Compare to your yeast’s published range: landing well below it usually means a stuck or incomplete fermentation, not a recipe that was meant to finish sweet.

Efficiency — why a recipe hits (or misses) its OG

Efficiency is how much of the grain’s available sugar you actually extracted. It’s the single most useful number to log, because every system has its own consistent figure, and feeding the right one into your recipe software is what makes predicted OG match reality.

• Conversion efficiency — did the mash convert the starch (crush, temp, time, pH)
• Mash/lauter efficiency — conversion + how well you rinsed the sugar out (sparge)
• Brewhouse efficiency — the whole-system figure into the fermenter, after the boil and all losses. This is the one recipe software wants. Homebrew all-in-ones typically land 65–80 %.

How to calculate brewhouse efficiency

Work in points × litres, which are conserved (boiling concentrates gravity but doesn’t change total sugar):

Collected = (measured SG − 1) × 1000 × volume (L)
Possible  = Σ (grain kg × ~300 points·L/kg for base malt)
Efficiency = Collected / Possible × 100

Worked example. 5.0 kg base malt; into the fermenter 21 L at 1.050:

• Collected = 50 × 21 = 1050
• Possible ≈ 5.0 × 300 = 1500
• Efficiency ≈ 1050 / 1500 = 70 %

Tell your software 70 % next time and it’ll predict your OG closely.

What it actually looks like over 140 batches

The archive on this site is big enough to answer the question empirically, so here is what efficiency really does rather than what the textbook says it should. Running the calculation above across every batch with a recorded actual OG and actual volume:

Three things fall out of that, and none of them are what I expected before I ran the numbers.

The gap between two similar rigs is eight points. Owen and I both brew on electric all-in-ones, both mill our own grain, both use broadly the same process — and his median sits eight points below mine, consistently, across seventy-odd batches each. That is not a mistake either of us is making. It is what “every system has its own number” actually means, and it is large enough to move a beer out of style if you swap recipes without adjusting. When I brew one of Owen’s recipes at his stated efficiency, I overshoot. When he brews mine, he undershoots.

The spread on one rig is wider than the gap between rigs. My own middle half runs 68–79 %, and the full range is 56–92 %. Grist composition is most of that — a big crystal-and-roast bill extracts differently from a pale one, and my flat 300 points·L/kg assumption over-credits sugar additions and under-credits dark malts. But even allowing for that, a single measurement is not your efficiency. Take a median across a dozen batches.

It has not improved with experience. The first half of my archive medians 75 %; the second half medians 72 %. If anything it has drifted slightly down, which I read as batch sizes and grists getting more varied rather than technique getting worse. The useful lesson is that efficiency is a property of your system and recipe, not a skill score to push upward — chasing a higher number by over-milling or over-sparging costs you in tannin and haze long before it pays.

Method note: this uses a flat 300 points·L/kg for every fermentable, which is right for base malt and wrong for sugar (higher) and roast (lower). Absolute values are approximate. The spread and the between-rig gap are the real findings, and neither depends on that assumption. Two barrel-fill batches were excluded, where the recorded volume is the barrel total rather than the batch.

When to take readings

Always record the volume with each reading. Gravity alone doesn’t give efficiency — gravity × volume does.

Common issues

• OG under target — low efficiency (coarse crush, mash pH off, under-sparge) or too much volume. Tighten the crush, check pH, and set the recipe’s efficiency to what you actually get.
• OG over target — boiled down too far / low volume; top up with sanitised water to hit gravity and volume.
• FG too high — under-attenuation: under-pitch, cold, or stuck. Warm and rouse.
• Refractometer FG looks wrong — it is; alcohol skews it. Use a hydrometer.