Saturday, December 10, 2016

Sugars in whisky

There are small amounts of sugar in all whiskies. Scottish whiskies have some sugars dissolved from the oak cask and often some from optional caramel colouring (E150a). Total amount of sugars is quite low, usually well below 1 g/l, but in certain cases it is quite possible to reach a few grams per liter. The sweet aromas of whisky matured in refill bourbon or new oak casks mostly come from sweet aromatic vanillin and fruity esters, not from the sugars. However, sugars can have a significant role in the case of casks previously used for sweet wine or sweetened spirit.

Typical composition of a liquid E150a colouring
Sugar gets to whisky from the oak cask used in maturation and its possible previous contents, especially sweet wine, and often from optional caramel colouring. New make spirit does not contain significant amounts of any sugars and it is not legal to add sweeteners to Scottish whisky, Irish whisky or bourbon, unlike to most rums, other American whiskies, Canadian whiskies, cognac, armagnac, vodka and so on. However, caramel colouring is an allowed additive in Scotch whisky. The E150a colouring used in whiskies consists of mostly of water (30-50%), carbohydrates (50-70%) and ash (0-4%), depending on the manufacturer. The colour comes mostly from furans and furfurals, mainly 5-hydroxymethylfurfural (5-HMF). The carbohydrate composition varies a lot, but usually the amount of simple sugars (mono- and disaccharides) is about 10-20% of the total volume and about 20-40% of the total carbohydrates. Caramel colouring as such is not sweet but rather bitter, although it probably enchances the perception of vanillin and some other sweet notes while supressing some sulphury notes. The amount of caramel added to whisky is usually well below 1 g/l, but there is no legal upper limit, although 2g/l of E150a (0,2-0,4 g/l simple sugars) would render the whisky very dark. In one independent study the highest caramel solids amount found in a Scotch whisky was 0.97-1.10 g/l (simple sugars ~0.2-0.4 g/l) depending on the analysis method (I suspect L*******g, but no names) and the average amount was 0.278 g/l caramel (from 0.01 to 1.10 g/l). According to Valaer in 1940 there was a legal maximum of one pint of liquid caramel per barrel of whisky permitted by the British Customs and Excise Regulations (0.15% or about 1.5 g/l).
Caramel content of various whiskies (Boscolo et al 2002)

Another source of sugars in whisky is wood. Oak is composed of three main macromolecules; cellulose is a long-chain D-glucose-fiber. Hemicellulose is a mesh-like branched chain of 5-carbon sugars (xylose, arabinose, galactose, ribose, rhamnose) and glucose. Lignin is a complex matrix of polypropane and polyphenols. Cellulose breaks down very slowly in alcohol-water solution extracting very small amounts of glucose to whisky. Most sugars extracted from the cask come from hemicellulose, hence most of them are 5-carbon sugars. Arabinose, xylose and rhamnose taste about half as sweet as glucose, where as fructose is almost twice as sweet compared to glucose.

About 100 mg/l of solids (including sugars) is extracted from new charred oak casks during the first year of bourbon maturation and after 1-3 years typically significantly less and only as little as 4 mg/l/year extraction has been reported from old refill casks used for example for cognac (or Scotch) maturation. A toasted cask extracts considerably more sugars compared to a charred cask, but typically about 150-250 mg/l/year is reached for the first year(s) of maturation. For comparison, a 40 years old brandy matured in a toasted oak cask had 2 g/l sugars (averaging 50 mg/l/year), while a 30 years old cognac (probably mostly matured in a refill casks) had only 0.5 g/l (17 mg/l/year).

Sugars dissolve in water better than in ethanol, so for example 55% ethanol-water solution can extract almost two times more sugars from the wood compared to 70% abv solution. The difference between 40% abv and 15% abv seems to be just about significant, so the alcohol strength is probably important only for high proof spirits. High entry proof is likely to significantly diminish the sugar extraction. Toasting and especially charring of the oak lowers the amount of extractable sugars, converting them to for example furfurals ("caramel"). On the other hand, charring contributes to sweetness by increasing the formation, extraction and perception of other sweet-tasting compounds, especially vanillin.
Analysis of different spirits by GC-MS (Savchuk 2001) Note the high levels of vanillin from new charred casks.
Probably the most important factor considering the amount of sugars in Scotch is the previous content of the cask. The most usual type of cask, the ex-bourbon cask would impart minimal amounts of sugars, often well under 50 mg/l/year and overall well below 1 g/l even during a long maturation, usually in the range of few hundreds of milligrams per liter. The charred American oak cask used for bourbon is quite low on sugars to start with and the bourbon will extract a major part of the available sugars. A toasted cask previously used for cognac, armagnac, brandy of other non-sugared spirit could extract slightly more, but most likely still not over amounts 1 g/l during refill maturation.

Venezuelan rum, 40 g/l sugars
Rum can, and often does, contain added sugar or other sweeteners. Most sweetened rums are sugared at the bottling phase, but sometimes sugar (or molasses, cane juice, fruit etc) has been added already into the cask and some of that sugar is bound to get into a ex-rum refill Scotch. Up to 40 g/l sugar contents are commonly reported in rum (100 g/l in spiced rums), but the amount of sugar added directly to cask is likely to be in the range of 5-20 g/l.

Pedro Ximenez sherry,
 470 g/l sugars
Significant amounts of sugar can be extracted from a sweet wine cask. Sherry is by far the most used ex-wine cask in the Scottish whisky industry. Most sherries, such as fino, manzanilla, amontillado, palo cortado and most olorosos, are dry wines (0-5 g/l sugars) and add only little sugars to the cask, some might even extract more than impart. Most sweet blended sherries (cream, dulce, sweet olorosos) are often blended with sweet must prior to bottling, not in maturation casks. The sweetness perceived in the olorosos often comes from high amounts of glycerol, not as much from sugars. A notable exception is Pedro Ximenez wine, which nowadays is vinified almost exclusively in Montilla-Moriles. It is a sweet wine produced from late harvested dried PX grapes and it typically contains by law over 212 g/l sugars. It is quite sweet even compared to other sweet wines, such as port (usually 100-150 g/l), Sauternes (100-200 g/l), Tokaji (60-500 g/l) or icewine (180-320 g/l).

The Scotch whisky industry has preferred sweet sherries for seasoning of their whisky casks from the 19th century and while most modern sherries are quite dry, most ex-sherry whisky casks were used for (or seasoned with) sweet sherries. In the 1800s and the early 1900s sweet(ened) sherries were imported to Britain in shipping casks. Britain was a major importer of sherry and emptied shipping casks were often used for maturation of Scotch whisky. By the late 19th century sherry casks were used extensively in Scotland and the first attempts to rejuvenate or season used casks were made in the 1890s. After the sherry casks were used for Scotch whisky maturation, they were seasoned with sweet sherry or paxarette, sometimes using pressure to soak the wood with the wine. By the early 20th century the use of sherry casks was quite wide spread, for example in the late 1920s a big whisky blender Johnnie Walker matured all of their whiskies in sherry casks or sherry-treated casks.
Aberlour 18yo
sugars <2 g/l

Shipping casks were mostly Spanish coopered lightly toasted American oak (Quercus alba) butts or puncheons (often called bocoyes in Spain). In 1972 Manuel González Gordon stated that "in recent years some Spanish oak has been used, due principally to the difficulties of importing American timber, but its greater density and hardness and lower porosity make it less suitable than the American wood". In the 19th century and the early 20th century they were often used for fermentation prior to shipping. Oak influence was not that detrimental for sherry during fermentation period and the fermentation process was believed to extract some of the excess tannins and bitterness from the wood prior the use in a solera or shipping.

The fermentation phase added plenty of sugars to the oak. For example, a typical 1980s (arguably Spanish oak, but at least Spanish coopered oak) shipping cask previously used for fermentation extracted 2.25 g/l sugars to whisky during only 6 years of maturation (375 mg/l/year). On the other hand an American ammonia-treated cask without a previous fermentation use would extract 770 mg/l (154 mg/l/year) sugars from a fino-seasoned cask and even less 655 mg/l (133g/l/year) from a similar oloroso cask. According to the table below, the previous use as a fermentation cask is very significant in the terms of sugar extraction.
JM Philp 1989
Ardbeg Dark Cove
4 g/l sugars
The most important factor considering the sugar extraction is probably the sweetest liquid to occupy the cask previously. According to Manuel González Gordon "large solera butts which have been used for years in a solera sometimes gain up to 25 kilos" in weight. In case of fermenting must or sweet wine, the refill cask can extract at least up to about 400 mg/l/year sugars. A typical sherry treatment used for Scotch whisky casks used in the 1890-1970s was made with about 35 litres of paxarette or sweet sherry per hogshead, but even with a relatively high pressure only a small part of the wine was absorbed into the cask. If we presume that as much as 5 litres of paxarette (375 g/l sugars) was absorbed into a hogshead and half of its sugars (just a figure from top of my hat) was extracted to the whisky, it would have resulted into 1875 g/ 220 litres = about 4.3 g/l of sugars, with a hypothetical gram or two from the cask itself could theoretically bring the sugar content up to 6 g/l during a very long maturation, say an average of 300 mg/l/year for 20 years. If were take a purely fictional solera butt used for an extremely sweet PX-wine (400 g/l) with 25 kg of PX wine absorbed into the cask and then used for whisky maturation, it could mean 10 kg of sugars in the cask and with a purely fictional extraction of about half of the sugars would result roughly to the amount of 9 g/l sugars into the whisky in an extreme(ly long) maturation from an extreme cask. It is probably safe to say that most spirits with over 10 g/l sugars contain additives and most spirits with over 2 g/l were matured in ex-sweet wine casks. Nowadays the sherry casks used in the whisky maturation are no longer used for fermentation or exposed to high-pressure paxarette/PX-treatment. The seasoning sherries are probably mostly medium oloroso/raya sherries with a sugar content in the range of 0-100 g/l, so a typical sugar extraction from that kind of cask could be in the range of 100-300 mg/l/year and diminishing through maturation, resulting to a maximum of a few g/l sugars into a modern whisky even during decades in a cask.

Alañón ME et al. Monosaccharide anhydrides, new markers of toasted oak wood used for ageing wines and distillates. Food Chem 2010;119;505-512
Blanco Gomis D et al. Evolution of sugars in cider brandy aged in oak barrels, a contribution to its characterization. J Agric Food Chem 2003;51;923-926
Boscolo M et al. Spectrophotometric determination of caramel content in spirits aged in oak casks. J AOAC Int 2002;85(3);744-750
Boudet AM et al. Biochemistry and molecular biology of lignification. New Phytol 199;129;203-236
Clyne J et al. The effect of cask charring on Scotch whisky maturation. Int J Food Sci Tech 1993;28;69-81
González Gordon M, Sherry. Cassell Ltd 1972
Haldane FF. Casks; their manufacture and treatment. J Inst Brew 1906;12(7);688-711
Hills, P. Appreciating whisky. Collins 2000
Kansallisarkisto, Helsinki. 
Liebmann AJ & Rosenblatt M. Changes in whisky while maturing. Ind Eng Chem 1943;35(9);994-1002
Martínez Montero C. Estudio de parámetros alternativos como indicatores del envejecimiento y de la calidad del brandy de Jerez. Thesis, Universidad de Cádiz, 2006
Martínez Montero C et al. Sugar contents of brandy de Jerez during aging. J Agric Food Chem 2005;53;1058-1064
Mosedale, JR. Effects of oak wood on the maturation of alcoholic beverages with particular reference to whisky. Forestry 1995; 68; 3; 203-230
Mosedale JR & Puech JL. Wood maturation of distilled beverages. Trends Food Sci Tech 1998;9;95-101
Piggott JR et al. Effects on scotch whisky composition and flavour of maturation in oak casks with varying histories. Int J Food Sci Tech 1993;28;303-318
Piggott, JR et al(ed). The Science and technology of whiskies. Longman 1989
Read J. Sherry and the sherry bodegas. Sotheby's 1988
Savchuk SA et al. Application of Chromatography and Spectometry to the Authentication of Alcoholic Beverages. J Anal Chem 2001;56(3);246-264
Tolman LM & Trescot TC. A study of the methods for the determination of esters, aldehydes and furfural in whisky. J Am Chem Soc 1906;28(11);1619-1630
Tsai PJ et al. Interactive role of color and antioxidant capacity in caramels. Food Res Int 2009;42;380-386
Valaer P & Frazier WH. Changes in whisky stored for four years. Ind Eng Chem 1936;Jan;92-105
Valaer P. Scotch whisky. Ind Eng Chem 1940;32(7);935-943
Valaer P. Foreign and domestic Rum. Ind Eng Chem 1937;Sep;988-1001
Williams JF. Rapid determination of alcohol in distilled spirits and of color in whisky. Ind Eng Chem 1926;Aug;841-843

Thursday, September 24, 2015

Scottish pot still variations

Scottish whisky distilleries accommodate a variety of different pot stills. Volumes vary from the 30963 litre wash still of Glenkinchie to the 2179 litre spirit still of Edradour (artisanal sma' stills excluded) and height from a couple of metres to the 8m tall Glenmorangie stills. Most are roughly onion shaped, but there are significant differences from the tear-shaped Bunnahabhain and Lagavulin to the slender Glenmorangie and Bruichladdich stills. Many stills have a wide bottom and a constricted neck, while several have a bulbous boil ball (aka reflux ball, Milton/Balvenie ball or bulge) in the middle section. Overall, the Scots are quite traditional with their stills, but occasional purifiers, still-neck coolers and a variety of condensers are still used. Most have converted to steam coil heating, while some still apply direct fire to the stills. In the history there has been a greater variety of stills, sadly more often to avoid taxation or speed up the distillation than to improve the distillate. In the 20th century the influence of North American distilling companies brought some interesting experiments with the reflux or Lomond stills, but the traditional pot still prevails in Scotland.
Tall Glenmorangie stills (photo: BBC)
Early distilling stills were usually quite small and direct-fired. There was not much of a commercial distilling until the 17th century. From 1644 the spirits in the United Kingdom were taxed by the gallon of proof spirit and the taxation did not really affect the distillation methods. Anyway, most distillers did not pay the taxes, as many distillers were just farmers saving the surplus grain for the winter as a spirit. In 1725 the malt tax was raised in Scotland to the English level. This caused riots and decreases in ale drinking, leading people to consume illegal spirits instead. In 1780 wine was taxed heavily and again the illegal distillers gained customers. A year later all private distilling was made illegal, though not to much effect in the Highlands. The use of mostly illicit stills affected the size of the stills used from the 18th century in the Highlands. They should be easy to transport or hide, so increasingly smaller stills were used as the Excise control became stricter in the late 18th century. 
Sma' still (decoration at Glenlivet, photo by mrtattiehead)

Due to rapid increase in illicit distilling, the crown allowed to licence over 20 gallon (91 litres) stills in the Highland area in 1784. As this did not have much effect, the limit was raised to 40 gallons (182 l) and all spirit exports from the Highlands were prohibited to England in 1785. At the same time the spirit duty was calculated according to the size of the still and the stills were assumed to be used seven times a week. This led to the use of very shallow 'Millar' pan stills from the 1790's, especially in the Lowlands, where the Excise surveillance was effective at least to some extent. The Lowland still was claimed to operate in 3 minutes, providing significant tax benefits, but extremely coarse spirits. The pan still was wide and shallow, usually just a few inches inches deep and usually just about over the minimum legal capacity. Additional upwards expanding coils or pipes were introduced into the still to maximize the evaporation at the bottom and condensation at the top.

Millar shallow stills (Forbes, 1948)
Lowland still sure was fast, but offered minimal reflux, copper contact and virtually no chance to cut the spirit correctly. So the coarse Lowland spirit often had to be rectified, ie redistilled or spiced heavily. Since the crown could not catch most of the illegal distillers and lost huge revenues because of the legal Millar stills, they decided to prohibit all stills under 2000 gallons (9092 l) in 1814. This effectively stopped pan distilling in the Lowlands, improving the quality, but put legal quality Highland distilling under ground until 1823, when taxation was eased considerably. In fact the over 40 gallons stills were legalized in 1816, but this had not much effect on the Highland stills or their desire to go legal.

Conversion to steam heating
*previously used steam jackets
**spirit still converted 2001
Another major improvement in efficiency was the use of steam, first directly on the pot or via a steam jacket, but later by indirect heating by steam coils. Direct steam heating for distillation was first implemented by dye manufacturers in the late 18th century with good heat economy. The first distillery to use direct steam heating of wash was English Mesly distillery in 1801. Indirect heating by steam jacket was used in Irish Roscea distillery in 1818, but the Excise Board rejected it because of difficulties in estimating the still capacity. The Coffey stills were heated by steam, but the pot stills were mostly direct-fired up to the mid 20th century. It was until 1887 Glenmorangie was the first Scottish distillery to use steam heating. Auchentoshan had steam jacketed heaters from early 1900s, but the majority of distilleries converted to indirect steam as late as the 1950-60s. In continental Europe and Scandinavia the steam heating became more widely used, especially after Savalle's invention of steam regulator in 1857.
Steam coils inside a spirit still
There are still Scottish distilleries using direct-fired stills; Glenfiddich, Glenfarclas and Macallan use direct-firing for both wash and spirit stills, while Springbank, Glen Garioch and Tobermory use direct-fired wash stills. Direct firing provides more temperature fluctuation both in the still and the contents. High temperatures can lead to burning and excess formation of furfural (nutty,burnt) or sulphurous and vegetal notes, especially if there is lots of sediment (dead yeast, grains) in the wash. To prevent this burning, direct heated wash stills have been installed with rummagers, ie rotating chains to keep the solids moving at the bottom of the stills.

Early 19th century was the golden era for still experiments. The still heads were adjusted in many ways; there were soap trays and stirrers to minimize frothing, steam coats to enable more even distillation and most importantly several means to enhance the reflux with boil balls, cooling condensers, refridgerators, trays and series of boiling chambers, which probably eventually led to the birth of the continuous stills by Adam, Fournier, Blumenthal, Stein and Coffey.
Balvenie spirit still with a boil ball

A common way to increase reflux was to shape the stills so that the rising vapours were cooled or exposed to lower pressure and copper. This was easily done by adding bulbs or balls in the still head. They were and are still common in the Speyside. Boil balls could be one reason for the perceived fruity and light "Speyside"-style of spirit, since they probably increase copper contact, reflux and ester formation, while decreasing the heavy oily and sulphury notes.

Still head cooler at Dalmore
Water cooling of the still head was probably first tested in late 18th century, the first patent is by Pontiflex from 1798, but the first Scottish distilleries to commercially cool their still heads were Hazelburn (1837), Dalmore (1839), (Ben) Nevis (1878), Fettercairn (1890) and Littlemill (1931). They used different methods for this, as Hazelburn had basically a tube condenser installed at the top of the still, where the spirit passed through pipes, which were cooled by water circulating in the condenser unit. Dalmore simply had a water jacket around the still head. Nevis, an extension of the Ben Nevis distillery, probably used a water jacket similar to the one in Dalmore, although there have been (unconfirmed) claims of a Lomond style cooled plate systems inside of the still. Littlemill had an early version of the Lomond still combined with water jacket coolers outside the still.
Nevis (Barnard, 1887)
Purifiers in the front left row, still head-cooler in the back

Intermediate condensing vessels between the still head and the worm were proposed already in 1736 by John Payne. The distilled wash or spirit is condensed inside a water jacketed purifier situated in the swan neck of the still and the heavier stuff is returned into the pot making the spirit lighter and higher in alcohol. Cooled intermediate condensers (water cooled purifiers) are still used at Glen Grant, Glen Spey and Strathmill distilleries.
Glen Grant uses purifiers on all stills
Glenugie purifier
Uncooled purifiers (basically a pipe descending from the lyne arm back to the still) are used in Ardbeg, Glenlossie and Talisker.

Ardbeg purifier (photo by
After the distillate leaves the still and there is no turning back via reflux or purifier, the spirit must be cooled. The traditional cooling method is a worm tub, consisting of a pipe wound into a spiral and immersed into a cold water tub. German chemist Christian Ehrenfried Weigel invented a worm tub in which the the cooling water was circulated to keep it cold (the Liebig condenser). The tube condenser with multiple straight copper pipes inside a cooled shell instead of a single coiled one in a tub was invented in 1825 by William Grimble. The shell and tube condenser provides more copper contact and is much more durable. A dozen Scottish distilleries still use worm tubs.

Disused worm at Auchentoshan
The Lomond still is a pot still with several horizontal (often perforated) plates inside the still neck. They became more popular in the Northern America and several bourbon and Canadian whisky distillers continue to use reflux stills, as they are often called over there.
Lomond stills at Glenburgie

The first commercial reflux still in Scotland was installed in Littlemill distillery in the early 1930s. American Duncan Thomas bought the distillery and apparently tried to produce light bodied Lowland whisky by two distillations with the reflux column and water jacketed cooler on the spirit still, instead of the traditional triple distillation previously used at the distillery. The double distillation with reflux column pot still was used at Littlemill until its closure in 1994. A similar experimentation was conducted in the 1950s by the Canadian company Hiram Walker & Sons. Their agenda was to produce a variety of different malt whiskies for blending purposes. To increase the control of the distillation reflux they used rotating rectifier plates, which could be turned to vertical or horizontal position for less or more reflux. The first such still was installed at Inverleven malt distillery situated inside the grain distillery Dumbarton in 1956. The resulting spirit was called Lomond whisky, hence the name Lomond still used later to describe all the Scottish reflux still whiskies. The experiment was successful, and the Canadians went on to install Lomond stills at Glenburgie (to produce Glencraig), Miltonduff (to produce Mosstowie) and Scapa. Scapa was different from the others, as it used a Lomond still in the wash distillation to render the spirit "sweeter and cleaner".
Lomond wash still at Scapa
Littlemill distillery owner Duncan Thomas founded the Loch Lomond distillery in 1966 with the American company Barton distillers to produce a yet wider variety of spirits (7) by altering the settings of the rectifier plates, but also the length and angle of the lyne arm by a peculiar turning telescope lyne arm. These adjustable lyne arms were also installed in Mosstowie and Glencraig later on.

The Scapa reflux wash still experiment was discontinued in 1971. The official reason for removing the rectifier plates was that the tube and shell condensers made the wash still plates futile. That sounds odd since the tube and shell condensers had been used for a century by then. The Lomond stills in Miltonduff and Glenburgie were mothballed in 1981, as the surplus of whisky resulted in rationalisation of the business and eventually forced Hiram Walker to sell out all their Scottish distilleries in the mid 1980s. Littlemill was mothballed in 1994 and destroyed in a fire in 2004. Duncan Thomas and Bartons sold Loch Lomond in 1971 to a private company, which keeps the reflux still distilling alive in Scotland.

A hybrid Holstein still from Germany
Copper pot still with a rectification column and a dephlegmator (cooling head) and a steel condenser
Barnard A. The whisky distilleries of the United Kingdom. Birlin ltd 1887
Forbes, RJ. Short history of the art of distillation. Brill 1948
Hume JR, Moss MS. The Making of Scotch Whisky
Morewood S. A philosophical and statistical history of the inventions and customs of ancient and modern nations in the manufacture and use of inebriating liquors. Longman 1838.
Nettleton, W. The manufacture of spirits. London 1893
Reaich, D. Influence of copper on malt whisky character. In Proceedings of 5th Aviemore Conference on malting, brewing & distilling. 1998
Ronde I. Malt whisky yearbook 2015. MagDig 2014.
Russell I. Whisky. Elsevier 2003.
Townsend B. Scotch missed. Angel's share 2000.
Townsend B. The lost distilleries of Ireland, Neil Wilson 1997
Udo M. The Scottish whisky distilleries. Black&White Publishing 2006.
Weir, RB. The history of the Distillers Company 1877-1939, Oxford Univ Press 1995

Friday, March 6, 2015

Scottish whisky mash bill

Malted barley is the main ingredient in Scotch malt whisky. Barley (Hordeum) is a member of grass family, just like most cereals in the world. There are two main types of barley: two-row and six-row. Both have six rows of seeds, but in the two-row varieties only two are fertile. The two-row varieties have usually plumper grains, less protein and husks and more fermentable sugars than the six-row varieties. The British malting barley varieties are and were almost exclusively of the two-rowed type, and the protein-rich six-row barley is mostly used for cattle feed. Barley can be sowed during the winter or in the spring, depending on the variety and the climate. Most distilling malts are made of spring barley, because they malt better with less dormancy issues. European brewers mostly use two-rowed barley, while Americans utilize mostly 6-rowed. Some varieties are naked, ie they do not have so much husks.

During the 17th and 18th century the main cereals grown in Scotland were barley, oats and rye. It is likely that all of them were used to produce distilled spirits (and whisky), although barley was probably favoured for its better enzyme activity. Additionally oats had too much husk for efficient mashing and rye tended to produce excess yeast growth. Mixed fermentations were probably often used, after all distilling was merely a way to preserve excess crop. Martin Martin in 1702 describes the practice on the Isle of Lewis: "The corn grown here is barley, oats and rye... Natives brew several sorts of liquors; as common Uisquebaugh, another called Trestarig, id est Aqua Vitae, three times distilled, which is strong and hot; a third sort is four times distilled Uisquebaugh-baul; id est Uisquebaugh...The Trestarig and Uisquebaugh-baul are made of oats". The oats were likely to produce a wash lower in alcohol and higher in congeners, so the process of three or four distillations makes sense. Multiple distillations and narrower cuts were most likely used to produce a palatable spirit, not so much to reach high alcohol strengths.

Chevallier barley was recently revived in Norwich by Dr Chris Ridout
During the 18th and 19th century the barley varieties used were local landraces, which were selected by the local farmers. In Scotland local landraces, probably most of the variations of Scotch Common, a two-row, narrow-eared, small-grained, early ripening variety, dominated until the 19th century. In colder climates, especially in the Northern Highlands some bere barley, a six-rowed landrace, was cultivated, but it was mostly used as cattle-feed. Most early 19th century English landrace varieties were probably close to Czech Hanna variety, a two-row, early ripening barley with a quite brittle straw. First generally successful variety in Britain was Cheval(l)ier, discovered in Suffolk, 1819. It spread across the country quite rapidly and most of the barley sown in the 1840's Britain was Chevalier, especially in the South. The other widely spread variety in England was Annat (1830) and in Scotland some of the Hanna varieties were still used in the late 19th century.

The price of maize and grain whisky (Weir 1995)
Malted barley was the main ingredient in both pot-still and column still whiskies in the 1830-40s, although column distillers began adding unmalted barley into the mash. In 1848 adjuncts, such as molasses, treacle and sugar were permitted in whisky production (1847 for beer) and further tax cuts made it possible for especially column distillers to produce whisky from a variety of  raw materials, mainly from cheap American maize. Maize was about 30% cheaper than barley at the time and as the production in the USA increased during late 19th century, it quickly became the major ingredient of column still whisky. By 1877 the average grain whisky mash-bill was: maize 77%, barley 20%,oats 2,5%, wheat 0,4%, rye 0,02%, malted barley 0,14%. The grain for column distillers was bought predominantly from abroad, less than 1% of the total grains and 2-6% of the barley were sourced from UK at the time. Oats, rich in fibers, were in mainly to improve the draining. The role of rye was to propagate yeast, while barley provided the enzymes. The cheap maize, efficient continuous Coffey still and the crisis in wine/cognac industry helped the boom of grain whisky in late 19th century. Brewers turned to foreign barley in the late 19th century mainly because of better quality of the Danish malt, not so much for the price. The  Irish distillers used a mixed grain bill from the late 19th century, usually consisting of barley malt (30-50%), unmalted barley (30-40%), oats (20-30%), rye (3-6%) and wheat (5-10%).

The pot-still malt whisky distillers were more traditional, even superstitious in their grain purchases compared to brewers and grain distillers. They used mostly local barley throughout the 19th century, although during the periods of bad harvests and/or higher whisky demands, as during the 1890's whisky boom, they were forced to use some foreign barley, mostly from Denmark, the Baltic and Morocco. There were other significant reasons for pot distillers to use local barley: They were trying to ban the grain distillers from marketing grain spirits as whisky or Scotch and proposed that only whisky made from Scottish grain could be called Scotch. The local farmers also bought much of the draff (waste product of the first distillation, great cattle feed) from the distilleries, so both benefited from direct sales to each other without any mediators. One likely explanation for the local sourcing of barley is the traditional and frankly sometimes superstitious approach on any improvement or change of the process or the product.

Beaven 1947
In the beginning of 20th century the predominant varieties were Annat, Goldthorpe, Archer, Spratt and Chevallier. After the World War I hybrid selections began to be made. Spratt-Archer and Plumage-Archer were popular and together comprised of 80% of malting barley until the World War II. The selections were made mostly based on the yield per acreage and the carbohydrate yield of the malt, flavour was rarely discussed. The foreign barley entered the distilling malt markets in 1920s, sourced mainly from Denmark and Chile (probably 6-rowed), and to some extent from Romania, Tunisia, Canada and Australia. Danish and Australian were preferred of for their quality, but for trade protectionist reasons the distillers (lead by DCL) agreed to by preferentially Scottish and English barley.

In the 1920s the prices of cereals varied widely and grain whisky was made from various ingredients based on the world market. Barley from California and Canada (small 6-row), maize from US and Argentina, oats from Scotland and Canada, even Brazilian manioc were used. The column grain distillers used the cheapest available raw materials, for example the maize bought from USA was usually grade 3, while the US domestic distillers used grade 1-2 maize.

During the WW II, Danish Kenia was grown widely for its better yields, however it was not good for malting and after WW II Pioneer (Kenia x Austrian Tshermarks) and Proctor (Plumage-Archer x Kenia) dominated until 1960 with a acreage up to 70%, although DCL seems to have preferred Zephyr. Due to rapid growth in whisky production in the 1950s, more English and foreign barley was used. If six-row barley was used, the smaller grains were sold to the distillers (more enzymes) and the plumper grain (more yield) to the brewers. Golden Promise and Maris Otter were introduced in 1965. Golden Promise became the barley of choice for distillers for its yield and enzyme activity until 1980s and Maris Otter was the brewers' malt, allegedly for its flavour. In the 1980s German Triumph and its many hybrids (Corgi, Natasha, Optic, Prisma, Camarge) surpassed the Golden Promise for their better yield and some winter varieties were introduced in Southern Scotland (Melanie, Halcyon, Regina). After that many different varieties have been developed and the suitable varieties for distilling and brewing are declared annually by Institute of Brewing and Distilling.
Improvement of spirit yield (Russell, 2003)

Brookes 2005
During the 20th century the acreage yield of barley has increased rapidly. Archer gave less than 3 tonnes per hectare as modern barleys for distilling malt give up to over 8 tonnes per hectare. Also the alcohol yield has improved drastically, from 300 litres of pure alcohol per tonne of dry malt to about 460 lpa/t.

Recently old varieties have been revived, mostly due to growing craft beer movement, but also by some malt distilleries. Bere barley was used in whisky production in Highland Park until 1926 and has since been used mainly for bere bannocks, but also for malt whisky. Bere was 6-row barley variety originating probably from northern Scandinavia with long stem and rapid growth (therefore also called 90-day-barley). Michel Couvreur revived the bere whisky in 1985 when he used bere from Orkney to distill whisky at Edradour. Since then at least Arran, Springbank and Bruichladdich have released bere whisky.

Sadly, there are no scientific comparisons between the flavour of different barley varieties. The early malting varieties were proportionally higher in protein and fat but lower in carbohydrates. Steeping times were much longer, probably because lack of knowledge and to some extent because of dormancy-prone barley varieties. The germination times were longer and the temperatures in floor maltings were more uneven than in modern maltings. These differences most likely made the wort more prone to infections of wild yeast and lactobacilli, along with mostly longer fermentations and lower starting gravity. The consistency was probably more viscous due to greater proportion of betaglucans to alpha-amylases. So the wort was likely to have more husks, dead yeast, autolysis products, lactobacilli, oils, diacetyl, esters (from acids and alcohols) and thicker in consistency causing easier burning in the wash still. The result was likely to be oilier, more sulphury and fruity spirit with more higher alcohols (fusels) and more furfural (from the husks, providing nutty aroma), assuming that the other factors were kept constant. It is unlikely that there are considerable differences within the modern barley varieties in terms of distilling, since the specifications for malting barley are quite strict. However, there is proof that the change of barley variety also changes the lactobacilli flora in the distillery, which might have at least some effect to the spirit if long fermentation times are used. The practice of malting (floor malting/industrial malting) has probably greater effect on flavour than the barley variety.

Maize mill at Dumbarton (I.Hume)

Because the use of enzyme additions is prohibited in Scottish whisky, the grain distilleries continue to use about 10% of malted barley in their mashes to guarantee sufficient enzyme content for starch degradation. Maize from the USA dominated the Scotch grain whisky mash bill from 1860s until early 1980s, excluding the war years. Since then, wheat has surpassed it, mainly because the trade regulations and taxes, not so much because of the actual cost of raw materials. Actually the price of unmalted barley has been quite competitive against maize and wheat in recent decades, but the processing problems have steered most grain distillers to wheat, mainly from France in the 1980s, but lately predominantly from the UK, with a fraction imported from Germany and France. North British is the last Scottish distillery to use predominantly maize, as the other "maize distillery" Dumbarton was closed in 2002. Some maize is still used at least occasionally in most grain distilleries, most likely to minimize viscosity problems. Maize has lower viscosity of mash compared to wheat and malted barley, because of lower amounts of betaglucan and pentoses. Since 1990s the Scottish distillers' maize has been bought exclusively from southern France due to trade barriers of EU. Bakers and farmers prefer high nitrogen wheat, so there is not a serious competition over low-nitrogen grains and distillers' wheat does not carry a price premium.

Biernacka&Wardencki 2013
The differences between wheat and maize spirits are surprisingly clear, there are less differences between wheat, triticale and rye spirits. Maize spirits contain about ten times less higher alcohols than the other grain whiskies, but have proportionally higher levels of esters (fruity). The higher amount of pentoses in maize also contribute to the higher furfural (nutty) content. The greater variety and amount of higher alcohols in wheat spirit account to more harsh, spirity and solventy notes in young spirit, but a likely to develop to a variety of acetals and esters, with more fruity aromas. The popcorn aroma sometimes present in blended whisky is likely to come from too much feints from the malt whisky involved, not from the grain component. The mouthfeel of wheat whisky can be oilier or waxier due to more arabinoxylans, while corn whisky is usually described cleaner and shorter.

Biernacka&Wardencki 2013
Agu RC et al. Production of grain whisky and ethanol from wheat, maize and other cereals. J Inst Brew 2006;112;4;314-323
Beal, AD & Mottram DS. Compounds contributing to the characteristic aroma of malted barley. J Agric Food Chem 1994;42;2880-4
Biernacka P & Wardencki W. Volatiole composition of raw spirits of different botanical origin. J Inst Brew 2012;118;393-400
Beaven, ES. Barley. Duckworth 1947
Briggs, D. Malts and Malting, Blackie 1998
Bringhurst TA. Barley research in realtion to Scotch whisky production. J Inst Brew 2015;1;1-18
Brookes, P. Barley Breeding and development in the UK, an historical perspective. Brew Hist 2005;121;25-39
Bryce JH et al (ed). Distilled spirits: Production, technology and innovation. Nottingham Univ Press 2008
Campbell I (ed). Proceedings of the Third Aviemore Conference on Malting, Brewing and Distilling. Intitute of brewing, 1990.
Collins TS et al. Profiling of nonvolatiles in whiskeys. Food Chem 2014;163;186-196
Cramer ACJ et al. Analysis of volatile compounds from various types of barley cultivars. J Agric Food Chem 2005;53;7526-31
Dong L et al. Characterization of volatile aroma compounds in different brewing barley cultivars. Sci Food Agr 2014
Hoff S et al. Influence of barley varieties on wort quality and performance. J Agric Food Chem 2013;61;1968-76
Hornsey, IS. A History of Beer and Brewing. Royal Society of Chemistry 2003
Lloyd, WJW. Adjuncts. Centenary review. J Inst Brew 1986;92;336-345
Mallett, J. Malt. Brewer's Publications 2014.
Martin, M. A description of the Western Isles. 1702
Martin, P & Chang, X. Bere whisky- rediscovering the spirit of an old barley. Brew Dist Int 2008/4;6;41-43
Mather, RT & Willkie, RT. Distillers' grain manual. Seagram, 1942
Piggott JR et al (ed). The science and technology of whiskies. Longman 1989
Priest, FG & Campbell I (ed). Current developments in brewing and distilling. Aviemore conference 1982.
Robson, F. Cameronbridge, a distilling giant. Brewer Int 2001;1;4;16-19
Russell I (ed). Whisky, technology, production and marketing. Academic Press 2003
Weir, RB. History of the Distillers Company 1877-1939. Oxford 1995.

Tuesday, November 18, 2014

Chill filtration

Same whisky, chilled with a drop of water on the left
Most of the Scotch whisky in the market is filtered in some way before bottling. The methods vary from crude physical filtering (ie spirit is run through a cloth or a cellulose sheet) to temperature and pressure controlled nanofiltering systems. 

The main reasons for filtering the spirits are removing any solid components, improving the flavour and avoiding the haze formation later in the bottle or glass. 

Spirits have been chill filtered at least from the late 14th century. Russian vodka was traditionally rapidly cooled after distillation with ice. The excess water transformed into ice and most of the oils hardened on top of that. Filtering became common before the 17th century. The early filters were usually made of cloth (felt, cotton), but also of paper, sand and charcoal. Chemical purification was also used quite early on, just as with wine, mead or beer. Spirit cleansing additives of the 17th and 18th century included ash, potash, burnt wormwood, and the gentry also used milk, eggs, fresh black bread, soda and isinglass to produce top quality vodka. 

Charcoal in Jack Daniel's distillery (
Coal filtering was also widely used in USA and Canada in the early 19th century. In 1810's Kentucky "layers of white flannel, clean white sand and pulverised charcoal (about 20 inches thick layer) from good green wood such as sugar tree hickory" were used as a filter. The now famous Lincoln county process consisting of maple wood coal filtration at a thickness of 10 feet was invented in 1825 by Alfred Eaton and is still used by Jack Daniel's and George Dickel. Active charcoal was widely used from the start of the 20th century. The first written evidences of active charcoal filters are from 1785 from St Petersburg, Russia. Farmacist T.Y. Lowitz found that raw distillates became clearer and less harsh tasting after treatment with "raw coal".  The activated coal filtering was patented in 1901 in Austria and in 1907 in Russia and used widely ever since, especially in vodka production.

Sand and coal filter tanks in Zhitomir vodka distillery
The distillers filter their spirits for various reasons, most of which are based on traditional practices and empirical evidence, not much has been released as scientific papers. In Tennesee, the Lincoln County process is believed to reduce fusel oils and harsh flavours, resulting in a "less grainy" spirit, according to John Lunn, the master distiller of George Dickel. Most bourbons are filtered before and after aging with active charcoal, usually said to "mellow" the taste. In China, Japan and Taiwan the rice spirits are filtered to remove higher alcohols (fusels) to enhance the flavour. The vodka producers usually just try to remove any colour or taste to render the spirit "clean". Scotch whisky producers claim that their filtering does not affect the flavour and only prevents the haze formation in the bottle or glass. Actually, charcoal is not very efficient in removing oils from the spirit, at least at higher alcoholic strengths and the normal filtration of rice spirits removes only a small fraction of higher alcohols, although an extremely efficient nanofiltration can reduce higher alcohols up to 44%. Interestingly, the amounts of higher alcohols has quite dramatically dropped during the 20th century, which probably diminishes the clouding, but also must affect the flavour, both directly and through lesser high chain ester formation.

EU limit [g/l]Modern spirit, est.Analysis 1905Analysis 1966
Scotch malt10~ 11,5-3,52,7-3,9
Scotch blend10<10,5-1,91,7-2,2
Scotch grain10<10,5-3,00,7-1,3
Bourbon10~ 1,5
Rice spirits102-5


Fruit brandies2


Canadian whisky10

Irish whisky10

Amount of higher alcohols in different spirits [g/l]

The haze formation occurs when the long chain fatty esters become insoluble in lower temperatures or at lower alcoholic strengths. The most crucial esters in haze formation in whiskies are ethyl laureate (ethyl-dodecanoate), ethyl-palmitate (ethyl-hexadecanoate) and ethyl-9-hexadeanoate (ethyl-palmitoleate). Alcocol content of 45% abv is believed to be a crucial limit for these long chain esters to percipitate in room temperature. Other compounds responsible for the cloudiness are high molecular weight lipids, especially sitosterol beta-D-glucoside (probably cask-derived) and slightly ethanol-soluble lignins from the cask.

The haze-forming long chain esters are not very aromatic, ethyl laureate gives some floral, fruity and waxy aromas, as ethyl palmitate and ethyl palmitoleate mostly contribute to the mouthfeel (waxy, oily), although they are reported to give some aromas of coconut and fruits. Probably the more important factor is their ability to act as a surfactant and to enhance/suppress other aromas.

(Coal) filtering increases the formation of short esters and the conversion of aldehydes and ketones to alcohols, probably because of the slight oxiditation caused by the process. On the other hand, some (7%) ethyl acetate (pear-drops) was removed in a typical Scotch filtration, so the net effect remains unclear. The highly efficient nanofiltration and active coal filtration techniques used by brandy and vodka industries remove great deal of long esters and some terpenes. Even a light filtration removed all of the nerol (rose,lemongrass) from apricot brandy and the same probably happens in Scotch whisky filtration. Studies on Glenlivet malt whisky indicate that significant amount of gallic acid was removed in filtering, but the same study found no difference on nosing aroma.
Plate and frame filter in a bottling plant
A typical filtering method used in Scotch whisky industry is a plate and frame filtration with cellulose sheets, medium pressure and cool temperature. Depending on the source the temperature of whisky is lowered usually to +5 - +10C, although some apparently use even colder temperatures (down to -10C?). According to Glenmorangie the cold stabilisation time is 3 hours and apparently up to 24h in some other bottling plants. In the 1990s all of the filters used in Scotch whisky industry were of the plate and frame type. Some distillers have at least experimented with easier and cheaper membrane filters, but there is little information about whether they are in use at a larger scale. The pressure use in the plate and frame filters is usually between 20-60 psi (138-414 KPa), a higher pressure means faster but less complete filtration. The particle retention size is 5-7µm, at least in one bottling plant. Compared to the modern nanofilters at 10nm (0,001µm), that is quite crude.

Plate and frame filter (
Because the filtration processes are not uniform in the Scotch industry, it is not clear how much the process affects the flavour. Certainly you can change the flavour profile and the mouthfeel of an estery and oaky whisky with slow cold filtration. A too strict filtration would likely dimish waxy/oily mouthfeel and floral, fruity and oaky aromas. Whether the current filtering practices do that, is uncertain.

References and further reading.
Braus H et al. Isolation and identification of a sterol glucoside from whiskey. Agr Food Chem 1957;5(6);458-9
Da Porto C. Effects of chill filtration on the composition of grape spirit. Wein-Wissenschaft 2000;55(1);7-12
Duarte FC et al. Physicochemical and sensory changes in aged sugarcane spirit submitted to filtering with activated carbon filter. Ciênc Tecnol Aliment.,2012;32(3);471-477
Glaub, R et al. Effects of various filter systems on sensory quality of fruit brandies. Kleinbrennerei 1998;50 (1);6–12
Himmelstein L. The king of vodka. Harper 2010
Hsieh CW et al. Develop a novel method for removing fusel alcohols from rice spirits using nanofiltration. Food Sci 2010;75(2); 25-29
Ko W et al. Removal of higher fatty acid esters from Taiwanese rice-spirits by nanofiltration.  p353. In Distilled spirits, ed Walker GM et al NOttingham Univ Press 2012.
Lachenmeier D et al. Defining maximum levels of higher alcohols in alcoholic beverages and surrogate alcohol products. Reg Tox Pharm 2008;50;313-321
Miljic UD et al. The application of sheet filters in treatment of fruit brandy after cold stabilisation. Acta per tech 2002;44;87-93
Taylor AJ, Mottam DS. Flavour science: Recent developments, RCS 1996
Persson KM. Med kol och kolonn. Spiritus 2005;7
Piggott JR et ak. The science and technology of whiskies. Longman 1989
Pirie G et al. Membrane filtration of whisky. Food & Drink 2000; p9-13. IChemE 2000.
Pokhlebkin W. A history of vodka. Verso, 1992
Puskas V et al. Influence of cold stabilisation and chill membrane filtration on volatile compounds of apricot brandy. Food Bioprod Proc 2013;91;348-351
Schidrowitz P, Kaye F. The determination of higher alcohols in spirits. Analyst 1906;31;181-194
Singer DD. The analysis and composition of potable spirits. Analyst 1966;91;127-134
Wisniewski I. Filtering out. Whisky Magazine 2011;97;27
Malt Maniacs E-pistle: study: