*To whom correspondence should be addressed. Tel.: 45 35 283323; Fax: 45 35 283245; E-mail: lit@kvl.dk Lebensm.-Wiss. u.-Technol., 34, 469 } 477 (2001) Prediction of Sensory Texture Quality of Boiled Potatoes From Low-"eld H NMR of Raw Potatoes. The Role of Chemical Constituents L. G. Thygesen,* A. K. Thybo and S. B. Engelsen L. G. Thygesen & S. B. Engelsen: The Royal Veterinary and Agricultural University, Department of Dairy and Food Science, Rolighedsvej 30, DK-1958 Frederiksberg C (Denmark) A. K. Thybo: Danish Institute of Agricultural Sciences, Department of Horticulture, P.O. Box 102, DK-5792 Aarslev (Denmark) (Received January 4, 2001; accepted April 18, 2001) Texture is a sensory attribute of uppermost importance for the preference of potatoes. In an earlier study, it was shown that sensory texture attributes of boiled potatoes could be predicted, to some extent, by partial least squares regression of relaxation curves obtained from raw potato samples using pulsed low-xeld 1 H NMR. The aim of the present study was to understand the chemical background for the low-xeld NMR prediction of the sensory texture attributes of potatoes. Correlations between low-xeld NMR relaxation curves, sensory texture attributes and chemical constituents are presented. It is found that prediction of sensory texture quality of boiled potatoes from relaxation curves of raw potatoes is possible mainly because the NMR relaxation curves contain information on water content, which is inversely correlated to starch content. However, low-xeld NMR seems to ower additional information, which is not included in the chemical variables like starch content. Apparently, adhesiveness and springiness are slightly better described by low-xeld NMR than by chemical descriptors. This indicates that low-xeld NMR is able to probe certain mouth-feel variables due to its sensitivity to the state of moisture in the sample, a parameter it is not possible to measure by any other instrumental technique.  2001 Academic Press Keywords: potato; texture; sensory; low-"eld H NMR Introduction Texture is a sensory attribute of uppermost importance for the preference of potatoes. The variation between potato tubers is a well-known biological phenomenon often resulting in variations in the "nal product. This in#uences one of the main challenges in the food industry: to develop and produce products of high and uniform quality. As consumers have become more de- manding due to increased information and purchasing power, it is of high importance for the industry to be able to produce uniform products of high quality. For this reason the industry needs rapid on-line spectroscopic methods in order to (1) predict the optimal use of po- tatoes and (2) sort potatoes in given texture categories prior to processing in order to obtain a more uniform product. Low-"eld H NMR relaxation data has shown to be highly correlated to texture of various products due to the phase and compartmentalization sensitivity of this method. Examples include staling of bread (Engelsen et al., 2001; Seow & Teo, 1996), e!ect of storage time of cooked wild rice (Ruan et al., 1997), juiciness and tender- ness of porcine meat (Fjelkner-Modig & Tornberg, 1986), warmed-over #avour in porcine meat (Br+ndum et al., 2000) and sensory "brousness and toughness of frozen cod mince (Steen & Lambelet, 1997). However, a majority of the studies concentrate on monitoring the e!ects of a given process rather than study the interre- lation between texture and relaxation data. A previous study showed that sensory "rmness, springi- ness, graininess, mealiness, moistness and chewiness at- tributes of boiled potatoes were correlated to low-"eld NMR relaxation data of raw and boiled potatoes, with the best correlations found for raw potatoes (0.54}0.90 for raw potatoes, 0.52}0.75 for boiled, Thybo et al., 2000). The work presented here is a continuation of this pre- vious study, and the focus in this article is thus not to demonstrate that low-"eld NMR prediction of sensory texture of potatoes is possible to some extent, but to understand the chemical background for such predic- tions. To that end, this study will present correlations 0023-6438/01/070477 #09 $35.00/0 doi:10.1006/fstl.2001.0788  2001 Academic Press All articles available online at http://www.idealibrary.com on 469