Definition of the Sensory and Aesthetic Spaces of Dry White Wines with Aging Ability by Experienced Tasters

: The popular appreciation of dry white wines is most frequently directed to young wines. However, present consumption trends comprise the valorisation of aged dry white wines. Therefore, the present work was aimed at the sensory analysis of aged dry white wines to define their sensory space and to understand which factors drive their quality evaluation by experienced tasters (critics, oenologists and students). Individuals were asked to evaluate several synthetic and aesthetic attributes and to characterise the analytic sensory profile through a Check-All-That-Apply (CATA) methodology. The quality evaluations were differently correlated with wine synthetic parameters according to the taster group. For both critics and oenologists, overall quality scores were driven by persistence and complexity. Moreover, quality was also highly correlated with power for critics and with balance for oenologists. Quality scores were highly correlated with wine browning (absorbance at 420 nm) for critics. The tasting panel showed a homogeneous analytic description of aroma, taste and mouthfeel consistent with wine age. The different ages could be associated with a continuous sensory space characterised by a decreasing perception of freshness and an increase in mature and mellowed descriptors. All wines shared an austere in-mouth perception elicited by their acidity, saltiness, bitterness, smoothness and dryness. The age prediction showed that most tasters failed to guess the aged wines that were more than roughly 13 years old, indicating that tasters were not familiar with the sensory features of white wines from 17 to 46 years old. In conclusion, experienced tasters consistently described the sensory space and recognised the high quality of aged dry white wines. Education programs may use the defined sensory spaces according to aging and to expand the range of quality perception by consumers.


Introduction
In wines, the meaning of ageing is concerned with the changes in chemical and aromatic composition that affect their sensory characteristics [1].The most favourable type of aging in dry white wines is characterised by the loss of fresh fruitiness, and the development of aromatic nuances of honey, beeswax, straw, hay or nuts [2][3][4][5].The yellow colour of an old white wine tends to vary from straw to amber [3,6].Deviation from this favourable evolution is illustrated by the so-called atypical aging.In this process, the character of old wines is present in young wines leading to flavours of mothball, soap, rotten eggs, garlic or cooked vegetables, walnut/curry and bruised apple [5,7].The distinction from wines with proper aging ability is not clear-cut since these aroma descriptors may also be found in developed top-quality wines [8].Thus, the thin line between proper and atypical aging justifies the examination of the respective distinctive sensory profiles.
The concept of sensory space, defined as the characteristic sensory features of a product shared by a group of individuals [9], appears to be appropriate to characterise Beverages 2024, 10, 44 2 of 15 wines that are difficult to distinguish.This approach has been applied to the concepts of varietal wines [10][11][12][13][14], regional characterisation [15] and "green" wines [16].Concerning aging, most research has been applied to red wines' aging bouquet [17,18].Thus, the scarcity of research on aged white wines and the present market interest in these styles, justify an adequate definition of their sensory properties, including those with an aesthetic significance, characteristic of fine wines [19].
The approach to developing sensory conceptual spaces in wines includes three steps: (1) identification of the sensory concept, (2) perceptual evaluation of the sensory space and (3) sensory space description [20].The first step is performed using a mental descriptive questionnaire [21].This work aimed to assess and define the sensory space of aged white wines, including the so-called aesthetic attributes characteristic of fine wines.The novelty of this work was mainly in the utilization of a wide variety of grape varieties and vinification techniques and a range of vintages, aimed at defining an overall common sensory space for aged dry white wines.

Wine Samples
Eighteen commercial white wines of different ages and two commercial red and rosé wines, used as distractors, from different Portuguese regions were used in this study (Table 1).The white wines were between 1 and 47 years old, and the distractors were 47 (red) and 15 (rosé) years old.The producers kindly supplied the wine samples.Before sensory analysis, three experienced tasters of the laboratory staff tasted the wines to check occasional flaws.

Tasting Panels
The study used three different experienced tasting panels.The first panel (critics) consisted of 9 wine critics and sommeliers (4 women and 5 men), aged between 26 and 72 years (mean 52 years).The second panel (oenologists) consisted of 7 winemakers and oenology scholars (2 women and 5 men) aged between 28 and 61 years (mean 49 years).These participants had more than 5 years of professional experience in wine tasting.The third panel (students) consisted of 14 oenology students (8 women and 6 men) from the second year of the Vitis Vinifera Master of Viticulture and Oenology Engineering (Instituto Superior de Agronomia, ISA), aged 22 to 40 years (mean 27 years).Although without formal training sessions, all participants were familiar with flavour description and professional experience was considered more relevant than training for this study.Three sessions were held at the ISA Microbiology Laboratory.The participants were volunteers and were informed that the tasting would be directed to assessing aged white wines.Critics and oenologists tasted the wines on 14 June 2021.The student session was carried out on 9 July 2021.

Tasting Conditions
The tasting was divided into two parts, with 10 wines each.All samples were kept at a temperature of 19 ± 1 • C, and the tasting room had windows open for natural ventilation.The bottles were opened 30 min before pouring into transparent glasses [23] covered by glass Petri dishes.To limit carry-over effects and memory bias, samples were randomly distributed among tasters according to a Williams Latin square design.The tasting sheet was divided into two parts (Supplementary Figure S1).The first part consisted of a score sheet for synthetic descriptors [24].The second part corresponded to the taste/mouthfeel and aroma description using a Check-All-That-Apply (CATA) methodology.The attributes chosen for the CATA assessment were selected from a previous online survey [21].Tasters could choose a maximum of 5 descriptors in each taste/mouthfeel and aroma list.After the CATA, respondents were asked to rate the quality and liking by drawing a line on a 9 cm unstructured scale anchored at both limits.The questionnaire ended with a question about the predicted age of the wine, where tasters indicated the age that they would give to each wine.

Data Analysis
The scores of the synthetic/aesthetic attributes were compared using one-way ANOVA by means of the Chi-square test (χ 2 ) with α = 0.05, since the score distribution was not normal.Therefore, the correlations among the attributes were obtained by using the nonparametric Kendall coefficient (tau-b).Analysis of the CATA was performed qualitatively since the data violated the test assumptions (i.e., no more than 20% of the expected counts were less than five and all the individual expected counts were one or greater).Furthermore, given the small sample size, both χ 2 and alternatives tests had too low a statistical power to detect statistically significant differences.Correspondence analysis (CA) was performed based on the contingency tables of the descriptors quoted more than 10% of the time in at least one wine.Two cluster analyses were run, after data standardisation, using the Euclidean distance measure and the Ward.D clustering method, to obtain descriptor clusters (descriptors as rows and wines as columns) or wine clusters (wines as rows and descriptors as columns).Multiple Factorial Analysis (MFA) included the results of quality, colour (Abs 420 nm), age prediction and analytic descriptors.All analyses were performed using the using free statistical software Jamovi, version 1.8, www.jamovi.org(assessed on 30 October 2021).

Wine Synthetic and Aesthetic Scores
The comparison of the median scores given by all tasters to the synthetic and aesthetic attributes did not show differences among the wines (Table 2), reflecting the high individual variability in their assessment.The quality evaluation of the 20 tasted wines is shown in Figure 1, illustrating the unevenness of the scores.In certain wines, the scores varied by more than 8 in a maximum scale of 10.

Wine Synthetic and Aesthetic Scores
The comparison of the median scores given by all tasters to the synthetic and aesthetic attributes did not show differences among the wines (Table 2), reflecting the high individual variability in their assessment.The quality evaluation of the 20 tasted wines is shown in Figure 1, illustrating the unevenness of the scores.In certain wines, the scores varied by more than 8 in a maximum scale of 10.The younger white wines (WLi18 and WDo20) tended to be scored with lower values.The rosé and red distractor wines also tended to have lower classifications (Figure 1).This behaviour indicated that tasters penalised wines that were not consistent with aged white wines by previously being aware of the purpose of the tasting.Interestingly, wine WBe11 (2011 vintage) was also in the lower range of values, probably because its sensory features obtained by the tasting panel were consistent with a younger wine (see Section 3.2.1).

Figure 1.
Distribution of quality scores given by critics, oenologists and students (box plots: •, average; horizontal line, median; lower horizontal dash, 10% of the scores; higher horizontal dash, 90% of the scores; box, 25% and 50% of the scores; ♦, extreme scores).Quality (cm) Figure 1.Distribution of quality scores given by critics, oenologists and students (box plots: •, average; horizontal line, median; lower horizontal dash, 10% of the scores; higher horizontal dash, 90% of the scores; box, 25% and 50% of the scores; ♦, extreme scores).
The younger white wines (WLi18 and WDo20) tended to be scored with lower values.The rosé and red distractor wines also tended to have lower classifications (Figure 1).This behaviour indicated that tasters penalised wines that were not consistent with aged white wines by previously being aware of the purpose of the tasting.Interestingly, wine WBe11 (2011 vintage) was also in the lower range of values, probably because its sensory features obtained by the tasting panel were consistent with a younger wine (see Section 3.2.1).
The variability in quality scores by experienced or expert tasters has been documented [25], demonstrating that subjects have a strong idiosyncratic appraisal of aesthetic judgements.Indeed, the comparison between the panels (Figure 2) showed a different quality assessment.Ballester at al. [25] speculated that high correlation among quality judgements may be found among individuals with the same academic background (e.g., winemakers).In the present work, the different backgrounds of critics or the different years of experience (oenologists and students) might explain the differences in quality evaluation, together with a different utilization of the scoring scale.
The variability in quality scores by experienced or expert tasters has been documented [25], demonstrating that subjects have a strong idiosyncratic appraisal of aesthetic judgements.Indeed, the comparison between the panels (Figure 2) showed a different quality assessment.Ballester at al. [25] speculated that high correlation among quality judgements may be found among individuals with the same academic background (e.g., winemakers).In the present work, the different backgrounds of critics or the different years of experience (oenologists and students) might explain the differences in quality evaluation, together with a different utilization of the scoring scale.

Effect of Experience on Quality Prediction
To understand more deeply, which factors might underlie the different quality evaluations of all tasted wines, correlations among the synthetic/aesthetic parameters were computed.The scores of all synthetic and aesthetic attributes are listed in Supplementary Table S1.When the responses of all tasters were merged, quality was highly correlated with persistence (linger) and complexity (Table 3).However, each tasting cohort contributed differently to this correlation.Indeed, critics associated quality with linger, power and complexity.Oenologists mostly associated quality with the number of flavours, balance, linger and complexity.The correlations obtained with students were all less strong than those of the more experienced cohorts.Nevertheless, students showed higher correlations with balance and complexity.Overall, these results show that the evaluation of aged white wines is consistent with the definition of fine wines characterised by the valorization of aesthetic attributes such as complexity, persistence or balance [19].The higher correlations obtained with balance than with harmonious probably reflect a semantic preference for balance.This type of sensation may also be dealt with under the recently described concept of elegance [26], justifying the use of this term instead of harmony or balance in future studies.
Interestingly, quality was highly correlated with liking for critics and oenologists.These results demonstrate that, with experienced tasters, quality and liking are intrinsically related [27] and it is difficult to define if individual preferences are elicited by the aesthetic quality or if preferred wines must share higher aesthetic attributes, as hypothetised by Malfeito-Ferreira [28].

Effect of Experience on Quality Prediction
To understand more deeply, which factors might underlie the different quality evaluations of all tasted wines, correlations among the synthetic/aesthetic parameters were computed.The scores of all synthetic and aesthetic attributes are listed in Supplementary Table S1.When the responses of all tasters were merged, quality was highly correlated with persistence (linger) and complexity (Table 3).However, each tasting cohort contributed differently to this correlation.Indeed, critics associated quality with linger, power and complexity.Oenologists mostly associated quality with the number of flavours, balance, linger and complexity.The correlations obtained with students were all less strong than those of the more experienced cohorts.Nevertheless, students showed higher correlations with balance and complexity.Overall, these results show that the evaluation of aged white wines is consistent with the definition of fine wines characterised by the valorization of aesthetic attributes such as complexity, persistence or balance [19].The higher correlations obtained with balance than with harmonious probably reflect a semantic preference for balance.This type of sensation may also be dealt with under the recently described concept of elegance [26], justifying the use of this term instead of harmony or balance in future studies.Interestingly, quality was highly correlated with liking for critics and oenologists.These results demonstrate that, with experienced tasters, quality and liking are intrinsically related [27] and it is difficult to define if individual preferences are elicited by the aesthetic quality or if preferred wines must share higher aesthetic attributes, as hypothetised by Malfeito-Ferreira [28].

The Influence of Colour on the Quality of White Wines
Colour is a factor that can influence the quality perception when wines are tasted in transparent glasses.As widely recognised [29][30][31], experienced subjects rely on top-down mechanisms [32] to assess wine quality and colour is the first sensory feature to drive their responses.In particular, older white wines, with brownish colour, tend to be regarded as over-developed [33].In this work, tasters were not asked to rate the colour since the absorbance at 420 nm provided an objective measure of browning.The positive correlations between quality and colour (Abs 420 nm), only for white wines, are depicted in Figure 3.
transparent glasses.As widely recognised [29][30][31], experienced subjects rely on top-down mechanisms [32] to assess wine quality and colour is the first sensory feature to drive their responses.In particular, older white wines, with brownish colour, tend to be regarded as over-developed [33].In this work, tasters were not asked to rate the colour since the absorbance at 420 nm provided an objective measure of browning.The positive correlations between quality and colour (Abs 420 nm), only for white wines, are depicted in Figure 3.
The tasters were aware that were experiencing aged dry white wines.Therefore, some degree of browning would be expected to influence the results.Indeed, light yellow wines were considered young and tended to be less well rated.Quality median scores lower than 3.5 were only given by critics to the less brownish wines (WMo16, WLi18 and WDo20, see Table 1).Similarly, the highest browning values were not scored lower than 6 for quality by critics (WBa91, WPe00 and WPe08).A lower correlation was obtained with students, indicating that they were not as influenced by colour as critics and oenologists, which was probably explained by the higher level of experience of the professionals [34].

Analytic Descriptions
The wines were subjected to sensory analysis using CATA to obtain an analytical description of aroma, taste and mouthfeel properties.The results of the three tasting cohorts were pooled and the results are presented in two contingency tables (Supplementary Tables S2 and S3).The tasters checked all the available aroma (17) and taste/mouthfeel (12) attributes at least once.Interestingly, tasters did not report any additional descriptors related to occasional over-development, which would be expected since old wines were being tasted.A total of 11 aromas were used for the subsequent correspondence analysis (CA), where 10 were cited at least 10% of the time in one of the  The tasters were aware that were experiencing aged dry white wines.Therefore, some degree of browning would be expected to influence the results.Indeed, light yellow wines were considered young and tended to be less well rated.Quality median scores lower than 3.5 were only given by critics to the less brownish wines (WMo16, WLi18 and WDo20, see Table 1).Similarly, the highest browning values were not scored lower than 6 for quality by critics (WBa91, WPe00 and WPe08).A lower correlation was obtained with students, indicating that they were not as influenced by colour as critics and oenologists, which was probably explained by the higher level of experience of the professionals [34].

Analytic Descriptions
The wines were subjected to sensory analysis using CATA to obtain an analytical description of aroma, taste and mouthfeel properties.The results of the three tasting cohorts were pooled and the results are presented in two contingency tables (Supplementary Tables S2 and S3).The tasters checked all the available aroma (17) and taste/mouthfeel (12) attributes at least once.Interestingly, tasters did not report any additional descriptors related to occasional over-development, which would be expected since old wines were being tasted.A total of 11 aromas were used for the subsequent correspondence analysis (CA), where 10 were cited at least 10% of the time in one of the white wines and bruised apple was also included because of its relationship with possible oxidation.Therefore, straw, caramel, wet/flint stone and curry were present in white wines but were not included, while spicy and earthy were not used since are characteristic of red and rosé wines.Regarding taste and mouthfeel descriptors, astringency and roughness were not used since they share a low frequency of citation for all wines, as would be expected in white wines (<10%) [35].Sourness was also cited less than 10% of the time (Table S3) probably because it has an unpleasant connotation and tasters preferred to use the term acidity.
The CA, based on the contingency tables only for white wines, was run using the retained aroma, taste and mouthfeel descriptors (Figure 4).The Chi-square value of the CA was 675 with a high significance in the discrimination (df = 323, p <0.001).The taste and mouthfeel descriptors were not run separately because the p-value was > 0.05, indicating that wine discrimination was mostly due to aroma descriptors.Indeed, CATA might not have had discriminant power regarding taste and mouthfeel properties [36].The first two CA dimensions explained 76.79% of the variance, which showed an adequate discriminant power elicited by aroma descriptors.The in-mouth perception more distant from the center of the plot was sweetness.This could be explained by the effect of fruity and flowery aromas on the so-called phantom sweetness in dry wines [37].
The localization of the quality scores of the three panels as illustrative variables in the central zone of the plot indicates that quality evaluation did not contribute to the sensory analytic discrimination of the samples.
Proximity in the space cannot be interpreted as direct association between wines [38].For instance, the closer position of WMo16 than WDa12 to chamomile does not mean that it smells more like chamomile.Consequently, the map is not sufficient to draw a conclusion on the differences between products.The basic correct interpretation is that the farther out a wine lies on an axis, the more frequently that attribute is associated with that wine [38].Thus, the CA biplot shows a clear distinction of the younger wines (WDo20 and WLi18), in the left part of the quadrant, from the older wines in the opposed quadrant (WDa74, WPe08, WBa91, WPe00 and WDa03).The attributes placed closer to the center mean that are not exclusive of any wine (length, saltiness, bitterness and acidity).The overall sensory projections show that the samples adequately covered the sensory space correspondent to dry white wines of different ages.The localization of the quality scores of the three panels as illustrative variables in the central zone of the plot indicates that quality evaluation did not contribute to the sensory analytic discrimination of the samples.
Proximity in the space cannot be interpreted as direct association between wines [38].For instance, the closer position of WMo16 than WDa12 to chamomile does not mean that it smells more like chamomile.Consequently, the map is not sufficient to draw a conclusion on the differences between products.The basic correct interpretation is that the farther out a wine lies on an axis, the more frequently that attribute is associated with that wine [38].Thus, the CA biplot shows a clear distinction of the younger wines (WDo20 and WLi18), in the left part of the quadrant, from the older wines in the opposed quadrant (WDa74, WPe08, WBa91, WPe00 and WDa03).The attributes placed closer to the center mean that are not exclusive of any wine (length, saltiness, bitterness and acidity).The overall sensory projections show that the samples adequately covered the sensory space correspondent to dry white wines of different ages.
To understand the associations between the sensory descriptors, a cluster analysis was run and is depicted in Figure 5.The obtained clusters may be described consistently using a metaphor according to each dominant perception.Indeed, freshness may be used to encompass the descriptors sweetness, fresh fruit, floral, citrus and chamomile, typical of young wines.Another cluster, described as mature, included the aromas of honey and dried fruit linked to body, length and viscosity.A third cluster included only aroma attributes (kerosene, bruised apple, butter, oak and beeswax), gathered under the overall perception of mellowed wines [21].The sensations of acidity and other in-mouth perceptions (saltiness, bitterness, smoothness and dryness) may be described under the umbrella of austere, mostly used in the popular wine press (e.g., www.wineenthusiast.com/basics/drinks-terms-defined/austere-wine-meaning, accessed on 5 May 2024).The frequency of citation of the descriptors in each sensory cluster is given in Table 4.
to encompass the descriptors sweetness, fresh fruit, floral, citrus and chamomile, typical of young wines.Another cluster, described as mature, included the aromas of honey and dried fruit linked to body, length and viscosity.A third cluster included only aroma attributes (kerosene, bruised apple, butter, oak and beeswax), gathered under the overall perception of mellowed wines [21].The sensations of acidity and other in-mouth perceptions (saltiness, bitterness, smoothness and dryness) may be described under the umbrella of austere, mostly used in the popular wine press (e.g., www.wineenthusiast.com/basics/drinks-terms-defined/austere-wine-meaning,assessed on 5 May 2024).The frequency of citation of the descriptors in each sensory cluster is given in Table 4.

Predicted Age
The previous cluster analysis may also be applied to wines, according to their sensory description (Figure 6).The obtained four clusters corresponded to different wines' ages, as perceived by the tasters.The results show that in 11 wines, the prediction differed by less than 3 years, comprising wines with real ages up to 12.5 years (Table 5).The difference in the prediction was higher for the older wines, being remarkable in the case of WDa74, where the difference was roughly 36 years.The previous cluster analysis may also be applied to wines, according to their sensory description (Figure 6).The obtained four clusters corresponded to different wines' ages, as perceived by the tasters.The results show that in 11 wines, the prediction differed by less than 3 years, comprising wines with real ages up to 12.5 years (Table 5).The difference in the prediction was higher for the older wines, being remarkable in the case of WDa74, where the difference was roughly 36 years.The prediction may be explained by the sensory conceptual space of the wines.Indeed, when wines were grouped according to their predicted age and the frequency of citation of the clusters depicted in Figure 4, the output may be regarded as the sensory evolution of dry white wines during aging (Figure 7).As expected, this evolution was characterised by a decrease in the freshness perception.Yet, some fresh fruitiness may be kept [8], accompanied by an increase in the evolved wine attributes, comprising mature and mellowed perceptual spaces.Interestingly, the austere perception underlies all ages with a slight reduction with time.This behaviour is consistent with the metaphor "mellowed by aging" proposed during the initial conceptual definition [21].

Dryness
The prediction may be explained by the sensory conceptual space of the wines.Indeed, when wines were grouped according to their predicted age and the frequency of citation of the clusters depicted in Figure 4, the output may be regarded as the sensory evolution of dry white wines during aging (Figure 7).As expected, this evolution was characterised by a decrease in the freshness perception.Yet, some fresh fruitiness may be kept [8], accompanied by an increase in the evolved wine attributes, comprising mature and mellowed perceptual spaces.Interestingly, the austere perception underlies all ages with a slight reduction with time.This behaviour is consistent with the metaphor "mellowed by aging" proposed during the initial conceptual definition [21].

Overall Factors Affecting Quality Evaluation
The evaluation of the different factors that influence quality was performed by Multiple Factorial Analysis (MFA), using absorbance (420 nm), sensory descriptors and predicted age as variables.The first two components explained 68.7% of the variance (Supplementary Table S4).The variables contributing to the PCA are shown in Figure 8.The axes oppose the perception of freshness (left position) to the perceptions of mature (beeswax and oak) and mellowed (dried fruit, honey and viscosity) wines.In the middle of the plan appears acidity, indicating that this attribute does not distinguishes younger from older wines.Even so, the onset of browning is regarded as a negative event [39].Figure 7 also shows that quality could be related to an acceptable degree of browning (Abs 420 nm).Therefore, the dry white wines most valorised by the tasters shared sensory attributes typical of older wines, within a predicted age range.

Overall Factors Affecting Quality Evaluation
The evaluation of the different factors that influence quality was performed by Multiple Factorial Analysis (MFA), using absorbance (420 nm), sensory descriptors and predicted age as variables.The first two components explained 68.7% of the variance (Supplementary Table S4).The variables contributing to the PCA are shown in Figure 8.The axes oppose the perception of freshness (left position) to the perceptions of mature (beeswax and oak) and mellowed (dried fruit, honey and viscosity) wines.In the middle of the plan appears acidity, indicating that this attribute does not distinguishes younger from older wines.Even so, the onset of browning is regarded as a negative event [39].Figure 7 also shows that quality could be related to an acceptable degree of browning (Abs 420 nm).Therefore, the dry white wines most valorised by the tasters shared sensory attributes typical of older wines, within a predicted age range.
Figure 9 shows the discrimination of the dry white wines based on a priori grouping according to age perception.Wines were clearly separated, with younger samples in the lower left quadrant and mellowed wines in the lower right quadrant, while mature wines were placed in the higher quadrants.Table 6 summarises the quality, sensory and colour characteristics according to the predicted age groups.
Figure 9 shows the discrimination of the dry white wines based on a priori grouping according to age perception.Wines were clearly separated, with younger samples in the lower left quadrant and mellowed wines in the lower right quadrant, while mature wines were placed in the higher quadrants.Table 6 summarises the quality, sensory and colour characteristics according to the predicted age groups.Figure 9 shows the discrimination of the dry white wines based on a priori grouping according to age perception.Wines were clearly separated, with younger samples in the lower left quadrant and mellowed wines in the lower right quadrant, while mature wines were placed in the higher quadrants.Table 6 summarises the quality, sensory and colour characteristics according to the predicted age groups.

Limitations of the Study
The fact that each cohort had a small number of tasters limited the significance of results regarding the effect of experience on the evaluation of synthetic factors affecting the aesthetic properties of the wines.In addition, the large variability in responses regarding synthetic/aesthetic attributes could probably be reduced by extensive training to calibrate the responses and by outlier deletion.However, no attempt was made to remove outliers since Parr [40] stated that this variability is intrinsic to sensory analysis and "not and error in the machine".Nevertheless, the sensory analysis using CATA provided information that elicited a consistent definition of the aging sensory concept of dry white wines.The utilization of dark glasses would have reduced the top-down effects on sensory description but the intent was to check if brownish wines could be underscored, which did not occur.In addition, regarding the complex nature of wine ageing it would have been interesting to check the evolution of the sensory attributes with time after pouring wine into the glasses [41].
The experimental outputs obviously depended on the wines selected for the study.Moreover, the valorization of old wines might have been induced by previously informing tasters about the purpose of the research as a deliberate strategy to attract highly reputed professionals.Therefore, even if the tasted samples covered the sensory range from younger to older wines, it would have been useful to have had clearly substandard dark-brown oxidised wines to establish the boundaries between not-faulty/faulty wines, which appear to be easier to define in whites than in red wines [42].
Another limitation concerns the use of other grape varieties, or wines from international regions known for their aging potential [43], to cover sensory features different from those of Portuguese wines.

Figure 4 .
Figure 4. Correspondence analysis of aroma, taste and mouthfeel descriptors of dry white wines.The median quality scores given by critics, oenologists and students were added as illustrative variables.

Figure 5 .
Figure 5. Cluster dendrogram of sensory descriptors elicited by dry white wines of different ages.

Figure 5 .
Figure 5. Cluster dendrogram of sensory descriptors elicited by dry white wines of different ages.

Figure 6 .
Figure 6.Cluster dendrogram grouping dry white wines of different ages according to their sensory description.

Figure 6 .
Figure 6.Cluster dendrogram grouping dry white wines of different ages according to their sensory description.

Figure 7 .
Figure 7. Sensory characterisation of dry white wines grouped according to their predicted age.

Figure 7 .
Figure 7. Sensory characterisation of dry white wines grouped according to their predicted age.

Figure 8 .
Figure 8.Multiple Factorial Analysis of dry white wines displaying the quantitative variables related to colour (Abs 420 nm), predicted age, sensory attributes and quality.

Figure 9 .
Figure 9. Multiple Factorial Analysis of dry white wines grouped according to the age perception.

Figure 8 .
Figure 8.Multiple Factorial Analysis of dry white wines displaying the quantitative variables related to colour (Abs 420 nm), predicted age, sensory attributes and quality.

Figure 8 .
Figure 8.Multiple Factorial Analysis of dry white wines displaying the quantitative variables related to colour (Abs 420 nm), predicted age, sensory attributes and quality.

Figure 9 .
Figure 9. Multiple Factorial Analysis of dry white wines grouped according to the age perception.Figure 9. Multiple Factorial Analysis of dry white wines grouped according to the age perception.

Figure 9 .
Figure 9. Multiple Factorial Analysis of dry white wines grouped according to the age perception.Figure 9. Multiple Factorial Analysis of dry white wines grouped according to the age perception.

Table 1 .
Origin and absorbance (420 nm) of the tasted wines.
a State wine research center in Nelas, Dão region.b Blended wines with local grape varieties.c Barrique aging on lees.d Grape variety.

Table 2 .
Chi-square (χ 2 ) values of the scores given to synthetic and aesthetic attributes of all wines by the tasting panels (df, degrees of freedom; significant p-values < 0.05).

Table 2 .
Chi-square (χ²) values of the scores given to synthetic and aesthetic attributes of all wines by the tasting panels (df, degrees of freedom; significant p-values < 0.05).

Table 3 .
Correlation (r)between the median quality scores and the rest of the synthetic descriptors and liking.Bold numbers indicate strong correlations using Kendall's tau-b (>0.600).

Table 4 .
Frequency of citation of the perceptual overall clusters.

Table 4 .
Frequency of citation of the perceptual overall clusters.

Table 5 .
Average age prediction of the white wines reported by the tasting cohorts.

Table 5 .
Average age prediction of the white wines reported by the tasting cohorts.Mean value of the predicted age given by each member of the tasting panel. a

Table 6 .
Range of variation in the quality, sensory and colour characteristics according to predicted age wine groups a .Wines listed in Table5and values used in the Multiple Factorial Analysis depicted in Figures7 and 8. a