I recently stumbled across what seems to be solid evidence for some effects of terroir on wine. It is not new research, but for some reason it had managed to elude me, and was only brought to my attention by John Winthrop Haeger’s book Riesling Rediscovered. The research is published in  and  – see end of this post. As the latter reference is available on Google Books almost in its entirety, that was my main source of information and where I found the figure shown below.
Twenty-five different Riesling vineyard sites were studied: 12 in the Pfalz, and 13 in the Mosel, Ahr, Nahe and Rheinhessen. The substrates yielding the soils of these sites included limestone, sandstone, greywacke, basalt, slate, porphyry, and breccia from the rotliegend age. The vintages 2004, 2005, 2006 and 2007 were assessed 8 months after the harvest by panels of 13-20 trained judges. The grapes were picked at optimal ripeness, as determined by the estate owners, one portion from each site being vinified as normal by the estate, and one portion subject to a standard winemaking protocol that was the same for all sites. The wine made by the estate was evaluated in duplicate, while standard process wine was evaluated in triplicate.
The results of a discriminant analysis are shown below (click on the figure to get a larger version). If I understand the point of a discriminant analysis correctly, it here means that two functions of the flavour profiles, F1 and F2, were found such that wines of the different rock types cluster when F2 is plotted against F1. Thus, if you know the flavour profile of the wine, you could calculate F1 and F2 and stand a good chance of predicting the rock type by checking where the point lies in the left hand graph.Necessarily, there were many details omitted from the conference paper, so it is difficult to judge the quality of the research, but to me nothing stands out as being obviously flawed and I would assume the work is sound. As such, for me it is the first convincing evidence for the soil substrate type having an impact on flavour profile. All the other evidence I have seen has been non-blind and anecdotal, and whenever blind tasting has been used terroir differences seem to disappear, even for people who would generally be regarded as expert tasters.
One interesting question about this research is to what extent the experimental protocol was essential in revealing the terroir effect, and to what extent it would also be clear to expert but non-trained tasters tasting blind. In other words, now necessary is the training and the discriminant analysis in uncovering the important of terroir? Also, more generally, I wonder how applicable the results are to other grapes and regions.
See my previous posts on the subject to understand my mildly sceptical attitude to terroir. I think I would still describe my attitude in the same way, but it has certainly just shifted considerably in the positive direction. However, in the introduction to the article I summarise here, it is made clear that terroir is hugely important in marketing, as a unique selling proposition, and as point of interest for punters. Marketing was the driver for this research, and whenever marketing is involved my bullshit detectors start twitching with heightened sensitivity.
 Andrea Bauer et al, “Authentication of Different Terroirs of German Riesling Applying Sensory and Flavor Analysis”, in “Progress in authentication of food and wine”, pp 131-151, American Chemical Society,Washington, DC, 2011
 A Bauer and U Fischer, “Factors causing sensory variation in Riesling wines from different Terroirs in Germany”, in “Oeno2011: Actes de colloques du 9e symposium international d’oenologie de Bordeaux”, pp 1087-1092, Dunod, 2011