Adapted proofing spreadsheet with specific gravity. Big thanks to Max Oppedahl and Greg Miller from UC Davis. [This spreadsheet needs updated again, because I used “in air” specific gravity data from TTB chart 6 against “in vacuum” data from TTB chart 4. Chart 6 also has vacuum data so I just need to sit down and convert it.
I also found a guru who makes a beautiful piece of software for this kind of work as well as a few related blending tasks.]
This post is going to evolve as I solve the riddles, but I should explain what we are trying to do and where we are at so far.
Distilleries have a lot of trouble accurately measuring ABV of various size samples unless they spend tens of tens of thousands on sophisticated densitometers. A robust alternative may be a $1500 multi use analytical balance, a cheap Chinese pycnometer, and some good data tables.
[What is becoming apparent is that cheap Chinese pycnometer are not remotely viable. The taper joining the bulb to the capillary tube lid is everything. If there is even slight rocking, a subtle pumping action will happen that disturbs the capillary tube. This will absolutely crush any changes of getting to four decimal places. No literature seems to discuss these finer points. Even cheap pycnometers with a calibration certificate to three decimals places have proven to suck.
The options may be to buy a quality pycnometer for a few hundred dollars or possibly to lapp the taper with diamond lapping compound. Lapping is very common in metal machining and lens grinding, but I haven’t yet found anyone discusses it in this context. I’m working on it.
The blue dye in the above photo may also help us prove quality. It can show us how well our taper fits and make it obvious how any rocking disturbs the capillary tube.
Accurately measuring alcohol content is startlingly hard. We need to make sure the challenges are factored into business models and not a punch in the face for any new business. Scientists of the past successfully used pycnometers to achieve amazing accuracy before U-tube densitometers were invented. The industry will benefit from relearning the finer points of the old process.]
The hydrometer has the limitation that you need enough of a sample to float the bulb which sometimes is multiple hundred milliliters. As an alternative, pycnometers can measure the density of samples smaller than even 5 ml. There are even pycnometer based ideas to measure the density of mere drops using special microscope slides.
Just like a hydrometer, pycnometer measurement is really sensitive to temperature. Sensitivity goes down as alcohol content increases. For ferments this will be significant, but for distilled products it will be far less so. This arises because of how the density of water and ethanol changes across temperatures. For low ABV beverages, A 1930’s Industrial and Engineering Chemistry article gives the rule of thumb that being inaccurate by a 0.1°C in temperature is equivalent to being off by 1 milligram (0.001). The rough math of that is that being off by 0.1°C is being off by 2/3 of a percentage point of ABV! Luckily, our analytical balance can weight to a tenth of an mg (0.0001).
To be accurate, we do not so much have to control temperature, (like many new distillers feel the need with hydrometers) as properly measure it to accurately correct values using the tables.
[I will have to take a 40% ABV off the shelf spirit, a calibrated pycnometer and the room temp then predict what the weight of the liquid will be and see how close I get to hitting it.]
This is all going to lead us to measuring the ABV of small, 25 ml, birectifier samples. For aged spirit samples with no sugar, we will be able to use the TTB’s method for measuring obscuration by evaporation. We then may be able to develop a protocol for measuring obscuration in distillates with high dissolved solids via pycnometer-to-pycnometer micro distillation. This will differ from what is frequently practiced by being smaller scale which will make the process go faster while hopefully also gaining accuracy.
A pycnometer differs from a volumetric flask in that its ground stopper has a capillary tube built in so that when filled, excess liquid runs over. The accuracy of a scribed line and interpreting a meniscus become none issues. Cheap ones can be had for $10 while exotic ones get pricey and some are vacuum jacketed to insulate them and some have built in thermometers. If a pycnometer claims a number like 100 ml, that is ball park, so the first thing you have to do is calibrate it yourself and record the information. Measurements are temperature sensitive so pycnometers are often handled with plastic tweezers.
An analytical balance is an incredibly sensitive scale. They break easily because of their neurotic precision so the best way to think of them is that they are only for adding significant digits to things you already roughly know the weight of. This means you should never weigh your watch because you probably don’t know the weight. If it was beyond the balance’s capacity, which is often quite low, it would damage the linearity. When we weigh a 100 ml pycnometer we already know the glass itself tares to roughly 20 g and 100 ml of water is roughly 100 grams. What we are then doing is turning that 120 grams into 119.9997. If you accept this principle you can get results for $1500 that even tens years ago used to cost $7,000+. Arroyo’s balance likely cost a fortune and took twenty times longer to operate than ours today. A tenth of a milligram balance is necessary for measuring ABV because a one percentage point change in ABV effects the density by 0.0015±. A weight to calibrate a balance of 0.0001 precision can cost $200.
To remove air bubbles from the pycnometers, as described in the TTB videos, we are going to use an ultra sonic cleaner. Its action rapidly de-gasses the liquid. This device will ultimately also be used to clean all the glassware used with the birectifier.
When we extend these techniques beyond the pycnometer to obscuration, we are going to use an oven (simple kitchen toaster oven) to evaporate a precisely known volume of sample. To keep this dry while it cools before weighing on the balance, we are going to use an old fashioned glass desiccator ($30). The TTB gives protocol for assuming the density of the residue and then factoring its effect on obscuration. It is very fast and can rapidly happen in triplicate. Need be, an Auber PID could be used to control the oven. All of these skills will transfer to a protocol where we measure the essential oil yield of botanicals to help us scale botanical charges for gin distillation.
In the long run, we will be aspiring to take essential oil yield information and use the birectifier to create small scale botanical concentrates of known oil content to prototype gins and slowly scale them up. The robot being built to automate the birectifier is being designed to scale from Luckow & Arroyo’s standard protocols to something that can also handle gin prototyping.
If you have any relevant ideas, unique experiences, criticism, or data tables that you really enjoy, please send them my way.