Members Log-in
01 July 1998
The classification of malt whisky
By
Charles MacLean
Traditional Classification. Classification was not a problem in the old days: the division, for fiscal and other purposes was simply Highland/Lowland (ie distilleries above and below the Highland Line, an imaginary frontier introduced in 1784, which stretched approximately from Dumbarton to Dundee).
By the late 19th Century three further 'Whisky Regions' were recognised: Campbeltown, Islay and Glenlivet - the latter approximating to our Speyside. This simple division was all that was required by blenders who divide Highland malts into 'Top', 'First', 'Second' and 'Third Class' for blending purposes - broadly speaking the dozen 'Top Class' malts (all Speysides) being used as 'top dressings' in a blend, and the 'Third Class' malts tending to be considered as useful 'fillers'. But with the rise in interest in single malts during the 1980s, distillery owners, consumers and writers began to look more closely at regional classifications. Especially they we - were interested in ways in which individual regions might be considered to bestow regional 'styles' or 'character' to the malts made there. Professor RJS McDowell (The Whiskies of Scotland) had divided the Highlands into 'The Glenlivets and their like', 'Dufftown', 'Northern' and 'Island' as early as 1968, but it was not until Wallace Milroy (Malt Whisky Almanac, 1986) that sub-division really got underway, quickly brought to geographical sophistication by Michael Jackson (The World Guide to Malt Whisky, 1987).
Milroy divided the Highland Region into Northern, Southern, Eastern, Western, Speyside, Islands and Orkney. Jackson followed this, but called the 'Southern Highlands' 'The Midlands', and subclassified Speyside according to its mam rivers, viz: the Findhorn, the Lossie, the Upper Spey, the Lower Spey, the Livet, the Fiddich and the Dullan, Strathisla, the Bogie and the Deveron. A simplified version of this classification of Speyside by rivers has long been used by The Society Spey, Lossie, Deveron and Findhorn. But is it really helpful? As Tim Fiddler says, "... I am not keen on the obsession with river valleys. Process water almost invariably comes from springs, not always adjacent to the distillery". Inter alia he proposes an interesting new classification for Speysides.
A new classification for Speyside by Tim Fiddler. "Classification begins with geography; coincidence of characteristics is merely a bonus ... It has to be admitted that Speyside has style on its side. It also has numbers, and the trouble is that as a classification it is unwieldy. Even if the malts of the Lossie, Deveron, Findhorn and Nairn valleys are subtracted, that still leaves 36 malts. A closer examination is not as easy as it is in, say Islay, with a sensible eight. For the purposes of differentiation and comparison it helps to have groups of moderate, proportionate size. Although 'Greater Speyside' could be considered to extend over a thousand square miles, the distilleries are huddled together in groups. Two reasonably-sized groups can be formed from those on the banks of the Spey; I call these the 'Banffshire Bank' and the 'Morayshire Bank'. The towns in the district provide another four: 'Rothes', 'Dufftown', 'Elgin' and 'Keith'. There is a group of 'Coastal' distilleries, and a group of 'Up-land' distilleries further inland. The groups vary in size from five to nine, and are therefore in proportion to other regional groupings. I find that this enhances my discernment and appreciation of the wide variety of Speyside malts."
I find Mr Fiddler's grouping reasonable and useful. Should the Society adopt it for its own classification, in the Bottling Lists? Chivas' Century of Malts. I am writing this with my head full of the fumes of one hundred malts. Earlier this week, Chivas Brothers invited half a dozen wine and spirits writers to a presentation of the 100 whiskies which go into the vatted malt, Chivas' Century. For this event, and following the procedure he used in creating the vatting, their Master Blender, Colin Scott, arranged his malts geographically as follows: North Speyside (28 malts, including those from Elgin, Keith, Rothes and 'The Coast') , South Speyside (25 malts, including the products of Dufftown, 'The Banffshire' and 'Morayshire Banks' and 'Up-land Speyside'), North Highlands (14 malts, including the North-eastern malts), South Highlands (15 malts, including those from the West, South, South-east and Central Highlands) and 'The Rest' (18 malts from Campbeltown, Lowlands, Islands, Islay. These were actually vatted separately in the creation of Century).
Although this arrangement was geographical, it was possible to detect family resemblances, even in broad districts such as the North and South Highlands and Speyside. The Southern Highlanders were marginally heavier, fruitier and more intense than their heathery northern cousins, while the Northern Speysides were firmer, sweeter and more aromatic than the Southern Speyside malts, which we generally found more cereal-like.
Clearly, classification by character, style or flavour is more useful to the consumer than mere geographical grouping, and although regional characteristics are familiar to us, we all know how difficult it is to place some malts - especially when they are drawn from a single cask. Classification by Clustering.
The third approach to classification abandons geography altogether, and seeks to group malt whiskies by aroma/flavour alone. Dr Wishart, a designer of statistical software, has used a statistical method known as 'cluster analysis' to classify malt whiskies. Although originally developed for studies in biological taxonomy, cluster analysis can also be used for market analysis. Dr Wishart's classification, which I have attached to this article, is provisional and on-going. He is keen to have your comments on his findings to date, so please let us know your vIews and we will pass them on to him. How it works is this. Dr Wishart analysed the descriptive terms used in eight current books to describe 85 readily available single malts m proprietary bottlings at around 10 years old. A vocabulary of some 800 aromatic and taste descriptors was compiled. These words were then bundled into a number of flavour/aroma groups: sweet, peaty, smoky, medicinal, honied, spicy, sherried, nutty, cerealy, fruity, floral. Each of the 85 malts was 'consensus coded' (2 where a majority of authors agreed, 1 where a minority agreed, o otherwise) according to the number of times a descriptor was applied to it. Using his Clustan software, Dr Wishart then classified the 85 malts into 'clusters', each having broadly similar taste characteristics.
The result is what is called a 'hierarchical classification tree' in which the 85 malts have been ordered and classed into a kind of taxonomy of malt whisky based on their flavours and aromas. Dr Wishart then examined - somewhat arbitrarily - the division of this tree into 10 groups of whiskies plus one singleton (see attached list). Although you may be surprised to find, for example, Knockando and Glen Grant clustered with The Macallan and Springbank, or Glenkinchie lumped with Highland Park, the methodology is interesting and the findings potentially of great value to the consumer who continually asks: "If I like Clynelish (etc) what others will I like?" But to obtain more meaningful clusters, the language of whisky tasting must be more rigorous, the descriptors more narrowly defined. I wrote to Dr Wishart when I first heard of his research pointing out the variations - even contradictions and inaccuracies - to be found in the tasting notes supplied in the eight books, including my own, which provided him with his vocabulary. I wonder what his clusters would look like if he used the highly imaginative descriptors employed by the Society Tasting Panel?
Unfiltered on Facebook