The HBD Palexperiment - Lab Analysis (Part II)


The HBD Palexperiment - Lab Analysis (Part II)

by Louis K. Bonham (Brewing Techniques - Vol. 7, No.2)

Louis Bonham concludes his two-part series analyzing the HBD Palexperiment beers. This time: final gravity, pH, CO2 and headspace air, and alcohol content.

In the previous installment of this column, I discussed the methods used to test 35 of the beers involved in the HBD Palexperiment for IBUs and bacterial contamination, as well as the results of those tests. This article concludes the report of our testing with a description of the final gravity, pH, CO2 and headspace air, and alcohol testing. Look for more on the sensory evaluation of the beer in a future issue of BT.

As reported previously, space limitations prevent us from publishing all the raw data here, but it is available on the BrewingTechniques website along with Part I of the results at http://brewingtechniques.com/library/backissues/issue7.1/bonham.html. A downloadable version of an Excel spreadsheet is also available on-line at the website.

Final Gravity Testing

Methodology: To assay final gravity, we used the ASBC protocol for determining final gravity by pycnometer (1). A pyenometer (right) is simply a bottle designed to contain a very precise amount of liquid at a specific temperature.

Before the testing, we cleaned the two pycnometers with a chromic acid solution, followed by a distilled water rinse and then a rinse with alcohol. When the pycnometers were completely dry, we weighed each one using a Cynmar milligram balance (Cynmar Corporation, Carlinville, Illinois) and recorded their dry weight.

Each pycnometer was then filled with distilled water that had been chilled to approximately 64 °F (18 °C). After filling, we carefully put the thermometer/cap in place, taking care that there were no bubbles in the pycnometer. As the temperature of the water rose to 68 °F (20 °C), the water in the pycnometer expanded slightly, and a few drops were expelled through the pycnometer’s sidearm. When the fluid in the pycnometer reached exactly 68 °F, we put the sidearm cap in place and carefully dried the outside of the pycnometer with alcohol and paper towels. We then weighed the water-filled pycnometer to the nearest milligram and calculated the weight of the 68 °F water that filled each pycnometer.

For the beer testing, we first rinsed the inside of the pycnometer with approximately 20 mL of the beer being tested, and then filled the pycnometer with degassed slightly chilled beer. Replacing the thermometer/cap, we waited for the beer to rise to 68 °F, immediately removing the small amounts of beer that were expelled through the sidearm using the tip of a paper towel. If any bubbles formed inside the pycnometer, we carefully removed the thermometer and added a few milliliters of beer with a transfer pipette. When the temperature of the beer read exactly 68 °F and there were no bubbles in the pycnometer, we put the sidearm cap in place and used alcohol to carefully clean and dry the outside of the pycnometer. We weighed the filled pycnometer on the milligram balance and recorded its weight, then subtracted the dry weight of the pycnometer to obtain the weight of the 68 °F beer that filled the pycnometer. From the weight of the beer and the water, we could calculate the specific gravity of the beer as follows:

beer weight ÷ water weight = specific gravity of the beer

with the results rounded to the nearest 0.0001.

Results: Obviously, this procedure is a bit more involved than simply using a hydrometer, but it is very accurate. Indeed, it is interesting to note the distinctions between the final gravities reported by the participants (presumably measured using a hydrometer) and the gravities that we measured. Although many of the participants reported values that were very close to those we measured, others reported values that were significantly different. Some of these differences could have resulted from errors in reading the hydrometers (or, of course, errors on our part during the testing), but I suspect that many of the differences in this regard can probably be traced to the use of inaccurate hydrometers (2).

Louis Bonham is an attorney with Holmes & Bonham, L.L.P., in Houston, Texas, specializing in federal litigation and intellectual property law. He is a member of the Foam Rangers homebrew club with 10 years of home-brewing experience. He is a Quill and Tankard award-winner for brewing writing and the organizer of the new Masters Championship of Amateur Brewing (MCAB) national homebrew competition. He can be contacted at lkbonham@hbd.org.

As I wrote last issue, the experiment’s organizers asked the participating brewers to generate a postboil kettle volume of 5 gallons, but a number of brewers did not do so. As with the IBU data, these brewers’ results would skew the results for some comparison purposes, and so I have calculated and included in my lab results an “adjusted” final gravity based on adding or subtracting the appropriate amount of water from the postboil wort. (These adjustments typically amounted to less than 0.001.) The final gravities ranged from a low of 1.0084 to a high of 1.0233, with an average final gravity of about 1.0156.

pH Testing

Methodology/results: To assay pH, we used a Corning 140 bench meter (Corning Inc., Corning, New York) with a Corning general purpose combination electrode and an independent temperature probe. We calibrated the meter using fresh standard solutions of pH 4.01 and 7.01, and we rinsed the electrode with distilled water between tests. The pH levels ranged from a low of 4.29 to a high of 5.01, with most of the beers in the 4.40–4.65 range.

Carbon Dioxide and Headspace Air Testing

Methodology: We analyzed CO2 content and headspace air using a Zahm & Nagel Series 7000 Air Tester (Zahm & Nagel, Buffalo, New York) owned by the Saint Arnold Brewing Company in Houston, Texas. After mounting a bottle of beer in the tester, the operator pushes down the two handles and locks them into place, which creates a seal and pushes a gasketed needle through the bottle cap. (This also secures the bottle into the testing unit.) When the cap is perforated, the pressure inside the bottle is released into a very small chamber equipped with a pressure gauge. After recording the pressure in pounds per square inch, the operator opens a valve that allows the gas and foam to bubble up through a calibrated reservoir of sodium hydroxide (caustic soda). The sodium hydroxide absorbs the CO2, leaving only the head-space air in the calibrated portion of the reservoir. The entire device is vigorously shaken 20 times to produce more CO2 and foam (this flushes any remaining headspace air from the bottle), and the testing unit is inverted twice to facilitate the absorption of all the CO2. At this point, the operator reads and records the amount of gas present in the calibrated reservoir — the headspace air.

After testing, we checked the temperature of each beer with a laboratory thermometer. (The beers were all kept in the same ice chest and were at 40 °F [4.5 °C]) before the testing began, and all were still within 2 °F [<1 °C] of this temperature after being tested.) We refreshed the sodium hydroxide solution after every 10 beers or whenever CO2 absorption appeared to be getting sluggish.

Results: We determined volumes of CO2 in solution by using the ASBC CO2 Chart (ASBC Method Beer-13). Values ranged from 1.65 to 4.27 volumes, with the majority of beers in the appropriate 2–2.5 volumes range. Headspace air levels ranged from a low of 1.15 mL to a few beers that exceeded 13 mL. (Typically, beers with this extreme level of headspace air had been dramatically underfilled.) Not surprisingly, the force-carbonated beers (approximately a quarter of the beers in the test) generally had significantly lower headspace air levels than those that were bottle-conditioned. (See the box “Fill Levels and Carbonation Levels,” next page.)

How did the Palexperiment beers stack up to commercial standards in this area? As with avoiding contamination, as amateurs we’ve got a very long way to go to catch the pros. The staling effects of headspace air are well documented (3) and as a result, commercial operations generally view 1 mL of headspace air in a standard 12-oz bottle as an absolute maximum. (With modern bottling lines, headspace air levels of 0.15 mL are commonplace.) Although a very small number of Palexperiment beers did achieve headspace air levels below 2 mL, virtually all of the beers had headspace air levels that far exceeded the acceptable commercial maximum, and with one questionable exception,* none of the beers even approached the 0.5 mL levels that George Fix reports are obtainable with amateur force-carbonating equipment (and no small amount of practice) (4).

*One beer did measure a scant 0.3 mL of head-space air. This beer was, however, extremely undercarbonated, so I seriously doubt that all of its headspace air was expelled into the tester. The lowest figure that I can confidently report was a respectable 1.15 mL.

The air levels we encountered — which I suspect are typical for home-produced beer — may not present a significant problem if the beer is stored cold and consumed soon after bottling. But over time, beer with this amount of air will stale much more quickly than most commercial beers, and such staling can occur in a matter of days if the beer is ever subjected to warm temperatures.

Alcohol Testing

Methodology: The industry standard method of testing alcohol content involves distilling off all the alcohol in a 100-mL sample of beer, diluting the condensate to 100 mL with distilled water, determining the specific gravity of this solution, and then using an ASBC chart to get the proper alcohol by weight or volume figures (ASBC Methods Beer-4A and 4B).† This is a time-consuming process, however, and there is no way we could have performed precision distillations of 35 beers in any sort of timely fashion.

Because alcohol has a markedly different refractive index than water, it has long been recognized that the alcohol content of a beer can be determined by measuring its refractive index and specific gravity and then using various formulas to calculate the alcohol content (5).‡ Although this method is not as precise as the distillation method, it is quick and fairly easy to perform. Our methodology was fairly straightforward. Using an Abbe 3 bench refractometer (Spectronic Instruments, Rochester, New York), we simply measured the refractive index of a degassed sample of the beer, and then plugged that value and the measured final gravity of the beer into a formula that calculated the alcohol content.

The formula I used is derived from the Berglund, Emlington, and Rasmussen regression equations cited in DeClerck (5). This equation, like that in the ASBC protocol, calls for refractometer measurements to be in Refractometer Scale Units (also called Zeiss Units), which is an older scale used in immersion refractometers. I was able to convert these equations to the more commonly used Refractive Index (RI) scale by using an equation derived by Siebert (6). (Readings from refractometers calibrated in Brix [percentage sucrose] can easily be converted to RI values using tables from any number of chemistry reference works, such as those found in the CRC Handbook [7].) Also, the original equations typically give results in terms of alcohol by weight. Because most craft and amateur brewers are used to thinking in terms of alcohol by volume, I have converted the equations to give results in this format. The formula I used is

†Because this process involves the distillation of an alcoholic beverage, federal law requires that you get the permission of the Bureau of Alcohol, Tobacco & Firearms to operate a laboratory still for this purpose. Such permission is not hard to get, however — I wrote the BATF and received a letter granting permission in very short order. In addition to distillation, other exotic methods exist to assay alcohol content (ASBC method Beer-4D [gas chromatography] and 4E [automatic alcohol analyzers], for example), but for obvious economic reasons these are rarely available to the amateur or craft brewer.

‡The ASBC recognizes the validity of this method (ASBC method Beer-4C), but because factors such as color and ash content can marginally affect such readings, the ASBC protocol requires the construction of a calibration curve based on samples of the particular beer that have been tested by distillation or other ASBC-approved methods.

A= [1017.5596 – (277.4 x SG) + (937.8135 X RI2) – (1305.1228 x RI)] x (SG ÷ 0.79)


A = percent alcohol by volume

SG = specific gravity of the sample (20 °C)

RI = refractive index of the sample (20 °C)

I have tested this formula with the values used in the examples given in DeClerck, Siebert’s article, and the ASBC manual, and it consistently reports the correct answer to within about ±0.1. (Look for more on using a refractometer in an upcoming issue of BT.)

Results: Alas, the best-laid plans oft go astray. Although some of the values we obtained matched up nicely with the rough estimates obtainable from using an O.G./F.G. table, many did not; indeed, some calculated values were off by an absurd amount. (See Table I, right.) So far I have not been able to diagnose the source of this apparent inaccuracy (my instruments check out as adequately accurate with reference solutions), but the problem could be related to inadequate degassing of the samples, operator error, miscalibration of the equipment, or a combination of these elements. Furthermore, as DeClerck points out, alcohol estimates based on O.G./F.G. measurements do not take into account the varying uses of sugars to grow yeast biomass and thus are inherently iffy, often being off by as much as ±50% (5). Nevertheless, although the typical results — alcohol levels slightly below what was anticipated — are consistent with the fact that pitching rates were apparently below optimal, the existence of a few clearly erroneous results may warrant taking these results with a grain of salt.

The Results Are In

The HBD Palexperiment was a truly noble venture, and I thank everyone who participated in it. As noted last time, the data generated from the HBD Palexperiment is a trove of information that I hope will help many amateur brewing scientists to develop their theories on various aspects of our craft, and I welcome suggestions or ideas regarding interpretations of the data.

For me, the widely disparate results in things like bitterness, final gravity, pH, and CO2 levels starkly demonstrate the limits to recipe duplication. Even with the same recipe and identical ingredients, key aspects of the finished product can vary widely due to things like equipment, water, temperature, and of course your brewing practices — results that were confirmed by the sensory evaluation panels.

The results of the contamination and headspace air tests should similarly serve as a reminder that while many of us can indeed brew exceptional beer, we should not get too cocky about the “superiority” of our techniques compared with those used in the production of commercial beer. There are still areas where few if any of us even approach the objective quality levels that professional brewers must meet every day.

Finally, the experiment also proved that you can not only contribute to the knowledge base of the amateur brewing community by participating in experiments like this, you can also have quite a bit of fun doing so.


(1)      This protocol (Beer-2A, Archived) has been removed from the current ASBC Methods of Analysis, but the ASBC recently reprinted it in its Laboratory Methods for Craft Brewers (ASBC, St. Paul, Minnesota, 1997).

(2)      Jim Martella and Paul Gatza, “How to Get the Most out of Your Brewing Instruments,” BrewingTechniques 5 (4), p. 50 (September 1997).

(3)      George and Laurie Fix, An Analysis of Brewing Techniques (Brewers Publications, Boulder, Colorado, 1997), pp. 137–139.

(4)      George J. Fix, “Control Beer Oxidation from Kettle to Bottle,” BrewingTechniques 6 (6), p. 63 (November/December 1998).

(5)      Jean DeClerck, A Textbook of Brewing (reprinted by the Siebel Institute of Technology, Chicago, 1994), vol. 2, pp. 427, 442–443.

(6)      K.J. Siebert, “Routine Use of a Programmable Calculator for Computing Alcohol, Real Extract, Original Gravity, and Calories in Beer,” Journal of the ASBC 38 (1), pp. 27, 29 (1980).

(7)      —, “Index of Refraction of Aqueous Solutions of Sucrose (Cane Sugar),” in CRC Handbook of Chemistry and Physics, 54th ed., Robert Weast, Ed. (CRC Press, Cleveland, Ohio, 1974), p. E-224.

(8)      Greg Noonan, New Brewing Lager Beer (Brewers Publications, Boulder, Colorado).

All contents copyright 2019 by MoreFlavor Inc. All rights reserved. No part of this document or the related files may be reproduced or transmitted in any form, by any means (electronic, photocopying, recording, or otherwise) without the prior written permission of the publisher.