Introducing a Standardized Sensory Analysis Method for Wine: A Methodology for the Recruitment, Selection, Training, and Monitoring of Assessors—Implementation on the Greek Variety “Agiorgitiko”

Abstract

The international wine trade plays a crucial role in the global economy, and an effective method for wine sensory analysis is essential. The International Organization of Vine and Wine (OIV) has issued a review document on wine sensory analysis, but further specialization and development remain indispensable. Our research team adopted standardized methodologies from the organoleptic evaluation of olive oil and the sensory analysis of milk and dairy products, which served as the foundation for creating a robust and repeatable methodology for the recruitment, selection, training, and monitoring of assessors for wine sensory analysis. A statistically validated method for wine sensory analysis was developed during this study. Simultaneously, a comparative study involving two distinct groups of candidate assessors was conducted, with one group trained in the organoleptic evaluation of olive oil and another untrained in this area, aiming to compare the effectiveness of untrained and trained assessors in further training for wine evaluation. Finally, the developed method was applied to sensorially evaluate 25 PDO Nemea wine products. Based on the results, it appears that the proposed method for recruiting, training, and selecting assessors is reliable and leads to sensory panels with excellent reproducibility. Furthermore, it establishes that assessors from an olive oil organoleptic evaluation panel have a higher likelihood of being selected as assessors for a wine sensory analysis panel.

1. Introduction

Wine plays a crucial role in international trade, with its global export value reaching EUR 36 billion in 2023—the second highest ever recorded. This trade primarily occurs between developed economies. Specifically, the top three wine-exporting countries by volume are Italy, Spain, and France. Additionally, significant wine exports come from Chile, Australia, South Africa, Germany, Portugal, Canada, the USA, Argentina, and New Zealand.

Conversely, the leading wine-importing nations, based on volume, are Germany, the United Kingdom, and the USA. Following closely are France, the Netherlands, Russia, Canada, Belgium, Portugal, China, Japan, Italy, and Sweden [1].

The importance of wine as a consumer product is evident from numerous studies conducted in the field of consumer research [2,3,4]. These studies often focus on the organoleptic characteristics of wine, evaluated by experienced assessors [5]. Interestingly, the process of comparing organoleptic evaluations between consumers and trained assessors extends beyond wine and applies to other products as well [6].

The significance of the organoleptic evaluation and sensory analysis of wine extends beyond the market; it also plays a crucial role in the wine’s production. The organoleptic characteristics of a wine can be influenced by several factors, including grape variety [7], grape-growing region, harvest conditions, soil type, and winemaking practices [8,9]. Numerous research publications have illustrated the use of sensory science in comparing winemaking parameters. For example, researchers have conducted studies on the organoleptic properties of wines with varying proanthocyanidin contents [10], as well as wines aged using oak chips [11,12] and during bottle aging [13]. Additionally, studies have successfully categorized wines based on their organoleptic characteristics related to specific growing regions [13,14]. Clearly, sensory analysis is useful for oenologists and winemakers, as winemaking practices [15,16] and cultivation regions significantly influence the organoleptic characteristics. Additionally, organoleptic analysis could become a useful tool during the final blending of wines to create the final product.

Numerous organoleptic test methods have been applied to wine research. More specifically, this includes triangular tests [16]; rank rating [8]; sensory descriptive profiling using discontinuous 3-point, 5-point, and 7-point scales [11,13]; the frequency of attribute citation method [7]; and sensory descriptive profiling using a continuous scale of 0–10 [10,17,18]. In research with discontinuous scales or with a continuous scale, the average is the most widespread methodology for calculating the value of each attribute [10,11,13,17]. These methods are mostly employed by experienced wine tasters who undergo further training during the research process. Although the training methodology is briefly described, detailed information is usually not provided. For the statistical validation of the citation frequency-based descriptive methods that have been developed, the reproducibility index (Ri) is mainly used [19,20,21,22]. This index utilizes the number of descriptors chosen by the judge in both replicates, as well as the number of descriptors chosen by the judge in the first and second replicates. The composition of the organoleptic panel significantly influences its performance. In order to explore this, researchers have conducted comparative studies involving different groups of assessors. Specifically, they have compared two distinct groups: professional sommeliers and trained assessors, both of which are engaged in descriptive analysis [23]. These studies extend beyond the realm of wine tasting. For instance, in the field of organoleptic analysis of complex odors, researchers have compared the performance of trained and untrained assessors [24]. Additionally, similar studies have been applied to food evaluation, where the performance of a group of trained assessors was compared to that of consumers [6].

The great importance of sensory analysis is demonstrated by the multitude of standards issued by the International Organization for Standardization (ISO) in this field. These standards describe general guidelines for the selection of assessors, using tests for color vision, ageusia, and anosmia, as well as for the training and monitoring of selected assessors [25]. In addition, instructions are provided for determining perception thresholds [26], the methods related to investigating the sensitivity of taste [27], and the methodology for training assessors in the detection and recognition of odors [28]. Simultaneously, the standards outline the subsequent stages after training, including instructions for establishing a sensory profile [29] and guidelines for monitoring the performance of a quantitative sensory panel [30]. These standards have been assimilated from the standards, official documents, or EU Regulations on the sensory analysis of specific products.

The International Organization of Vine and Wine (OIV) has created a document as a review of the sensory analysis of wine [31], in which many elements from the ISO standards have been assimilated. This document represents the consensus reached by the members of the group of experts called the Subcommission Methods of Analysis of the OIV and is a proposal for the selection and training of tasters for the organoleptic analysis of wine. It provides very detailed and precise instructions for a series of tests, including tests to detect disabilities and determine sensory awareness. The document covers various topics, such as describing odors, identifying defects in wine using solutions, and addressing the most common defects in wine. Additionally, it includes guidelines for the sensory evaluation of different types of tannins, as well as instructions for assessing both assessor and panel performance. Specifically, this document presents a list of the main visual, olfactory, tactile, and taste-altered characteristics of wines, which are intended for use in taster training. Although the document provides detailed guidance in some areas, such as the initial training, much of it offers general guidance, necessitating specialization and expansion. Furthermore, it draws a connection between the accreditation for sensory analysis of wines and related works conducted in the field of olive oil.

The International Olive Council (IOC) has published a series of standards on the sensory analysis of olive oil. These standards describe the method of organoleptic testing for virgin olive oil [32], and additionally, they provide very detailed guidelines for meeting the requirements of ISO 17025 for sensory testing laboratories for virgin olive oil [33]. The IOC has also issued a guide for the selection and training of tasters, which includes specific objectives for assessing candidate tasters and determining their acceptance or rejection [34]. Furthermore, internal quality control guidelines for sensory laboratories [35] cover detailed instructions and examples for evaluating individual tasters and panel performance. This evaluation involves using duplicates and calculating metrics, such as the precision number and deviation number, with specific acceptance criteria for each parameter. The accompanying standards include a general basic vocabulary [36] and a method for the organoleptic assessment of extra virgin olive oil, which is applied to the designation of the origin [37].

Further, the literature on sensory analysis in milk and milk products has also been particularly developed. For this product category, the International Organization for Standardization (ISO) has issued three standards. These standards provide general guidance for the recruitment, selection, training, and monitoring of assessors [38]. Additionally, they recommend methods for sensory evaluation for each category of product, proposing attributes for each product [39]. Furthermore, the ISO standards offer guidance on a method for evaluating compliance with product specifications, including instructions for creating a relevant test report [40]. In more detail, the standard on assessors [38] provides comprehensive tests for basic odor and taste recognition, as well as the training and evaluation of assessors across different product categories. Specific scoring targets are set for the selection or rejection of each assessor. The evaluation tests cover both olfactory and gustatory aspects, including positive attributes and defects.

In the realm of wine, the imperative to establish comparable panels and achieve methodological harmonization in wine sensory analysis has prompted research efforts. These studies aim to furnish a guide on selecting and training tasters, with a specific focus on the taste attributes of wines bearing a Protected Designation of Origin (PDO) rating [18]. The study provides meticulous instructions for training assessors in the utilization of a continuous scale, employing standards to assess certain taste parameters and their corresponding concentrations.

The purpose of the present study is to develop a standardized method for the organoleptic analysis of wine by employing visual, olfactory, and taste attributes. This method is based on the proposals of the OIV. This study not only implements and refines the specific OIV proposals but also extends them by incorporating standards from other domains, such as virgin olive oil, milk, and milk products, as well as drawing upon relevant research in the field of wine.

The present study outlines a detailed process for training, selecting, and monitoring assessors. It provides an in-depth description of the analysis method and a comprehensive quality control methodology for both the sensory team and individual assessors. The detailed procedures include descriptions of the solutions used, a defined scoring system for assessor selection, and specific performance acceptance criteria. The ultimate goal is to create a standardized method that can be easily applied and replicated by other sensory panels.

At the same time, the two groups of candidate assessors for the developed method are compared: an accredited group for the organoleptic evaluation of virgin olive oil, which has not received training in the organoleptic analysis of wine, and an untrained group of candidate assessors, with no experience in sensory analysis, known as the “virgin” group. The aim is to evaluate whether the groups involved in the organoleptic evaluation of virgin olive oil can serve as a pool of candidate assessors that can be leveraged for the organoleptic analyses of wine, potentially increasing their chances of final selection.

Finally, the developed method is applied to the analysis of 25 commercial wine PDO Nemea products made from the Agiorgitiko variety. The results are used to investigate the discrimination ability of the standardized sensory method. Additionally, the study explores whether there is any correlation between the retail price of Nemea products and specific organoleptic attributes.

The main purpose of this specific study is to offer a sensory analysis method that is easily applicable and produces repeatable results. This method will serve as a valuable tool for oenologists and winemakers during the wine production process and in the international trade of these products.

2. Materials and Methods

2.1. Wines

A total of 25 commercial Nemea wines purchased by the Nemea Winemakers Association were analyzed. The price of the wines ranged from EUR 5 to 30 per bottle. The vintages for these wines were 2015 and 2016.

For the reproducibility and monitoring tests of panel performance and taster performance, seven wine products were purchased from the market. These wines were produced from Greek grape varieties, specifically Moschofilero, Assyrtiko, Savvatiano, Malagouzia, Agiorgitiko, and Xinomavro. The vintages for these wines were 2017 and 2018.

2.2. Chemicals and Reagents

A total of 22 compounds were purchased from commercial sources: vanillin, 4-ethylphenol, cis-3-hexen-1-ol, and isoamyl acetate were purchased from Acros Organics (Geel, Belgium); caffeine was purchased from Alfa Aesar (Ward Hill, MA, USA); butyric acid, citric acid, sodium chloride, and tannic acid were purchased from Honeywell Fluka (Charlotte, NC, USA); 2,4,6-trichloroanisole (TCA) was purchased from HPC Standards (Borsdorf, Germany); potassium disulfite was purchased from Merck (Darmstadt, Germany); acetic acid and ethyl acetate were purchased from Penta (Prague, Czech Republic); and 2-phenyl ethanol, 4-ethyl guajacol, benzyl-acetate, citral, ethanal, monosodium L-glutamate, sucrose, and tartaric acid were purchased from Sigma Aldrich (St. Louis, MO, USA). Grape tannin and “Qualitan Grape” were purchased from Dolmar (La Rioja, Spain). D(+)-Glucose was purchased from Panreac Applichem (Barcelona, Spain). The list of the standards, Chemical Abstracts Service Number (CAS Registry Number), molecular formula, purity, and company are shown in

Table A1. List of standards, Chemical Abstracts Service Number (CAS Registry Number), molecular formula, relative formula mass (Mr), purity, and company.

a/a	Name	CAS	Molecular Formula	Mr	Purity	Company
1	2,4,6 trichloroanisole (TCA)	87-40-1	C7H5Cl3O	211.47	99.5%	HPC Standards (Borsdorf, Germany)
2	2-Phenylethanol	60-12-8	C8H120	122.16	99.3%	Sigma Aldrich (St. Louis, MO, USA)
3	4-ethylguajacol	2785-89-9	C9H1202	152.19	98.0%	Sigma Aldrich (St. Louis, MO, USA)
4	4-ethylphenol	123-07-9	C8H100	122.17	99.2%	Acros Organics (Geel, Belgium)
5	Acetic acid	64-19-7	C2H4O2	60.05	99.8%	Penta (Prague, Czech Republic)
6	Benzyl-acetate	140-11-4	C9H1002	150,17	99.9%	Sigma Aldrich (St. Louis, MO, USA)
7	Butyric acid	107-92-6	C4H8O2	88.11	99.9%	Honeywell Fluka (Charlotte, NC, USA)
8	Caffeine	58-08-2	C8H10N4O2	194.19	99.7%	Alfa Aesar (Ward Hill, MA, USA)
9	cis-3-hexen-1-ol	928-96-1	C6H12O	100.16	98.6%	Acros Organics (Geel, Belgium)
10	Citral	5392-40-5	C10H160	152.23	96.0%	Sigma Aldrich (St. Louis, MO, USA)
11	Citric acid	77-92-9	C6H8O7	192.12	98.0%	Honeywell Fluka (Charlotte, NC, USA)
12	D(+)-Glucose	50-99-7	C6H12O6	180.16	99.0%	Panreac Applichem (Barcelona, Spain)
13	Ethanal	64-17-5	C2H4O	44.05	99.0%	Sigma Aldrich (St. Louis, MO, USA)
14	Ethyl acetate	141-78-6	C4H8O2	88.11	99.7%	Penta (Prague, Czech Republic)
15	Ethyl caproate	123-66-0	C8H1602	144.21	99.7%	Acros Organics (Geel, Belgium)
16	Isoamyl acetate	123-92-2	C7H1402	130.19	99.0%	Acros Organics (Geel, Belgium)
17	Monosodium L-glutamate	142-47-2	C5H8NO4Na	187.13	98.0%	Sigma Aldrich (St. Louis, MO, USA)
18	Sodium chloride	7647-14-5	NaCl	58.44	99.5%	Honeywell Fluka (Charlotte, NC, USA)
19	Sucrose	57-50-1	C12H22O11	342.30	99.5%	Sigma Aldrich (St. Louis, MO, USA)
20	Potassium disulfite	16731-55-8	K2S2O5	222.33	99.0%	Merck (Darmstadt, Germany)
21	Tannic acid	1401-55-4	C76H52O46	1701.20	99.0%	Honeywell Fluka (Charlotte, NC, USA)
22	Tartaric acid	87-69-4	C4H6O6	150.09	99.5%	Sigma Aldrich (St. Louis, MO, USA)
23	Vanillin	121-33-5	C8H803	152.15	99.5%	Acros Organics (Geel, Belgium)
24	Whiskey lactone	39212-23-2	C9H16O2	156.22	99.4%	Sigma Aldrich (St. Louis, MO, USA)

.

2.3. Equipment and Software

For the present organoleptic tests, wine-tasting glasses were used, which complied with the requirements of ISO 3591 [41]. During the training, whenever deemed necessary, the specified glassware was black to hide the color of the sample. Watch glasses were used to close the rim of each glass.

The Fizz sensory software “Acquisition”, v.2.47B from (Biosystemes, Couternon, France) was employed to generate the assessor sheet, determine the order of sample presentation, and assign three-digit codes to each sample. After the tests, data entry for further processing was performed using Fizz Calculations software, v.2.47B (Biosystemes, Couternon, France).

2.4. Panelists

A total of 20 candidate assessors were recruited for basic training in sensory analysis. The goal was to create a sensory panel of 10 assessors. In order to achieve this, twice the number of individuals actually required to constitute the final panel [25] were recruited. The age range of the 20 candidates was between 25 and 80, with an average age of 41.4 years. Thirteen of the candidate assessors were female, and seven were male.

The candidate assessors were divided into two groups, each consisting of 10 people. Group A comprised those candidates who had received training in the organoleptic evaluation of virgin olive oil and participated in an accredited panel for evaluating olive oil by the Hellenic Accreditation System (E.SY.D.). In contrast, Group B consisted of candidates who had not received training in the organoleptic evaluation of olive oil. Importantly, neither Group A nor Group B candidates had prior experience in sensory wine analysis.

During the selection process, candidates completed an initial questionnaire regarding their interest in organoleptic testing and availability. The selection criteria included consideration of the candidates’ interest in products of vitivinicultural origin and their experience in sensory evaluation activities [31]. Additionally, factors such as availability, motivation, ability to communicate, good health, and the ability to isolate themselves from external stimuli were taken into account [25]. Data were obtained from a relevant questionnaire and personal interviews with each candidate assessor.

2.5. Training Process

In total, 28 training sessions were conducted over a period of 6 months. The training started with theoretical training, which was based on international standards [25]. The theoretical training encompassed a broad range of subject matter, including the types of assessors, panel leader responsibilities, recruitment methods, screening tests, applications of sensory profiling, and general test conditions.

2.5.1. Tests to Detect Organoleptic Inabilities

Candidate assessors were then tested for their ability to accurately perform organoleptic analysis by conducting an ageusia test for three basic flavors and astringent sensation, as well as an anosmia test using four volatile substances involved in wine aroma [25], as presented in detail in

Table A2. List of standards and concentrations for the ageusia and anosmia tests.

Taste or Odor	Material	Concentration in Water at Room Temperature g/L	Concentration in Ethanol at Room Temperature mg/L
Taste
Sweet	Sucrose	10
Acid	Citric acid	0.3
Bitter	Caffeine	0.3
Astringent	Tannic acid	1
Odor
Lemon, Fresh	Citral		1
Vanilla	Vanillin		1
Floral	Benzyl acetate		1
Tropical fruits	Isoamyl acetate		1

. For these compounds and concentrations, it is generally accepted that candidates with less than 80% successful results should not be selected as assessors.

Subsequently, tests for stimulus detection [25] were conducted, employing triangular tests [42] with three taste parameters and one odor parameter. The materials and concentrations utilized are detailed in the Appendix A (

Table A3. List of standards and concentrations for stimulus detection.

Material	Taste	Mass Concentration or Volume Fraction in Water at Room Temperature
Caffeine	Bitter	0.2 g/L
Citric acid	Sour	0.2 g/L
Sucrose	Sweet	1.3 g/L
(Z)-Hex-3-en-1-ol	Green, grassy, unripe	0.4 mL/L

).

The selection section for detecting inabilities in organoleptic tests was completed with a color blindness test, where each tester underwent evaluation using the Ishihara method.

2.5.2. Threshold of Perception

The perception threshold for each candidate assessor, as well as for the group as a whole regarding the five basic tastes, was determined using a method based on international standards [26]. Specifically, each taster was presented with three solutions by employing the three alternative forced choice (3-AFC) methodology. In this approach, one of the three solutions contained the substance to be examined dissolved in water, while the other two samples consisted of water alone. The taster’s task was to select the sample in which they perceived taste. This procedure was repeated for increasing concentrations. Subsequently, the perception threshold was calculated for each individual candidate assessor and for the entire group using the “Shortcut procedure using Best Estimate Threshold (BET)”. Notably, the perceptual threshold was assessed for the five basic tastes: sweet, sour, salty, bitter, and umami.

2.5.3. Tests to Determine Sensory Awareness

Tests to Detect a Stimulus

Tests to determine organoleptic perception were conducted. During the process of detecting taste stimuli, the five basic tastes—sweet, sour, salty, umami, and bitter—were thoroughly examined, along with astringency. These tests were carried out using aqueous solutions containing characteristic substances for each of the six flavors, employing triangular tests [31]. Detailed information regarding the substances and their concentrations are shown in

Table A4. List of standards and concentrations to detect a stimulus.

Taste	Material	Concentration in Water at Room Temperature
Sweet	Sucrose	10 g/L
Sour	Tartaric acid	0.5 g/L
Salty	Sodium chloride	2 g/L
Bitter	Caffeine	0.3 g/L
Umami	Monosodium L-glutamate	0.6 g/L
Astringent	Tannic acid	1 g/L

.

In addition, the process of identifying tastes was carried out. Solutions were created for the five basic tastes, with the substances and concentrations described in international standards [26]; these are shown in

Table A5. List of standards and concentrations for the identification of tastes.

Taste	Material	Concentration in Water at Room Temperature g/L
Sour	Citric acid	0.28
Bitter	Caffeine	0.195
Salty	Sodium chloride	1.19
Sweet	Sucrose	5.76
Umami	Monosodium L-glutamate	0.29

. Candidate assessors received a sample of each type and were given the necessary time to familiarize themselves with it. They were then presented with a series of the same materials, including one or two glasses containing water. More specifically, they were given 10 glasses corresponding to one of the five flavors, as well as 2 glasses of water. After each trial, the tasters had to identify the characteristic taste of each sample.

Discrimination Tests between the Levels of Intensity of a Stimulus—Ranking Tests

The next training session involved ranking tests, which assessed the discrimination ability of each candidate. These tests were applied to color, odor, and taste [31]. The CIELAB color coordinates and the substances for odor discrimination, taste discrimination, and astringency discrimination, along with the concentrations for each level, are presented in

Table A6. List of standards and concentrations for tests to discriminate between the intensity levels of a stimulus.

	Level 1	Level 2	Level 3	Level 4
Color discrimination for red wines
Chromatic co-ordinates a:	30	40	50	60
Chromatic co-ordinates L:	60	50	40	30
Odor discrimination
Acetic acid	0.10 g/L	0.30 g/L	0.50 g/L	0.70 g/L
2,4,6 trichloroanisole (TCA)	1 ng/L	3 ng/L	5 ng/L	7 ng/L
Ethyl acetate	10 mg/L	40 mg/L	70 mg/L	100 mg/L
4-ethylphenol	0.05 mg/L	0.15 mg/L	0.30 mg/L	0.45 mg/L
Taste discrimination
Sucrose	2 g/L	6 g/L	10 g/L	14 g/L
Tartaric acid	0.25 g/L	0.50 g/L	0.75 g/L	1.00 g/L
Sodium chloride	0.5 g/L	2.0 g/L	3.5 g/L	5 g/L
Caffeine	0.05 g/L	0.20 g/L	0.40 g/L	0.60 g/L
Monosodium L-glutamate	0.05 g/L	0.30 g/L	0.60 g/L	0.90 g/L
Astringency discrimination
Tannic acid	0.50 g/L	0.75 g/L	1.00 g/L	1.25 g/L

. During the tests, samples were presented to the candidates following a Latin square design. The evaluation of the ranking tests was carried out with a scoring system from 0 to 4 points [38], as presented in

Table A7. Gradation scale for ranking the tests.

Points	Result
4	Four correct
3	Adjacent switch
2	Two correct without adjacent switch
1	One correct
0	None correct

.

2.5.4. Tests of Aroma Description

A series of solutions was presented to the candidate assessors, each possessing a characteristic odor above the perception threshold. These selected odors were intentionally associated with the aromas found in wines. The tests were conducted using ISO 3591 black wine glasses, ensuring that the color of the solution remained hidden from the tester. During the testing process, the samples were presented individually, and each candidate was asked to record the odor they perceived. Following the candidate’s initial reaction, the instructor had the opportunity to discuss the sample further, exploring the candidate’s ability to describe the sensory stimulus. The solutions chosen and their basic aroma descriptions are detailed in

Table A8. Olfactory materials for the odor description test and marking schedule for odor recognition.

Material	Characteristic Odor 3 Points	2 Points	1 Point
Isoamyl acetate	Banana	Tropical fruits	Fruity
2-Phenyl ethanol	Rose	Jasmine	Floral
Caramel syrup	Caramel	Vanilla, malt, toasted	Sweet
cis-3-hexen-1-ol	Green grass	Green beans	Green vegetable
Butyric acid	Rancid butter	Parmesan cheese	Vomit, baby burp
Vanillin	Vanilla	Custard, dessert	Sweets, candy
Citral	Lemon	Orange	Citrus, fruity
Ethyl caproate	Apple	Pear	Fruity
Fresh orange juice	Orange	Lemon	Citrus, fruity
Acetic acid	Acetic acid	Vinegar	Sour/off wine

. The selected odors are commonly found in wines, including vanilla, caramel, and lemon [31]. Simultaneously, we considered the proposed aromatic compounds described in milk assessors training, which are also relevant to wine, during the aroma selection process [38]. The concentrations for the solution of each substance were based on international standards [28]. For those substances without relevant reports, a solution was created with a concentration 10 to 20 times higher than the perception threshold. In such cases, the trainer evaluated the solution and its odor intensity, and any necessary adjustments were made.

The candidates’ performance was scored as follows: 3 points for a correct identification or description of the most frequent association, 2 points for a description in general terms, 1 point for a distant association identification or description, and 0 points for a non-response or a completely incorrect response. Examples of the scoring method for the selected substances are presented in Table A8.

2.5.5. Evaluation of Training—Selection of Assessors

After the basic training, the candidate assessors were evaluated on three tests: basic taste recognition, ranking tests, and aroma recognition.

Basic Taste Recognition

Five solutions of the five basic tastes were prepared at selected concentrations, as shown in

Table A9. Materials and concentrations for basic taste recognition.

Taste	Material	Concentration in Water at Room Temperature g/L
Sweet	Sucrose	10.0
Salty	Sodium chloride	2.0
Sour	Citric acid	0.3
Bitter	Caffeine	0.3
Umami	Monosodium L-glutamate	0.6

. These five solutions, along with a water sample, were presented to the candidate assessors using three-digit random codes. The order of presentation was randomized but consistent for all testers. Each tester was then asked to identify the base taste represented by each solution. Candidates’ responses were scored on a scale of 1 to 6 points, as indicated in

Table A10. Gradation scale for taste recognition.

Points	Result
6	All correct
5	Four correct
4	Three correct
3	Two correct
2	One correct
1	None correct

.

Ranking Tests

The candidate assessors were presented with four groups of solutions. Each group consisted of four solutions with a characteristic taste or aroma. The testers were asked to rank each group’s four solutions in order of increasing intensity. The solutions were presented with three-digit codes and a Latin square design. The groups and concentrations of each solution are shown in

Table A11. Materials and concentrations for the ranking tests.

Taste	Material	Level 1	Level 2	Level 3	Level 4
Sweet	Sucrose	Water	5 g/L	10 g/L	15 g/L
Sour	Citric acid	Water	0.10 g/L	0.30 g/L	0.50 g/L
Salty	Sodium chloride	Water	1.0 g/L	1.5 g/L	2 g/L
Bitter	Caffeine	Water	0.10 g/L	0.30 g/L	0.50 g/L
Odor
Barrel	Wine spiked with Vanillin and Whiskey lactone mix (50–50% w/v)	Wine	0.25 mg/L	0.50 mg/L	0.75 mg/L

. The candidate assessors’ performance was scored from 0 to 4 points, as indicated in Table A7.

Odor Recognition

A series of nine solutions were presented to the candidate assessors—each with a characteristic odor associated with the aroma of wines—at a concentration above the perception threshold. The tests were carried out in ISO 3591 black glass wine glasses, ensuring that the color of the solution was not visible to the tester. The solutions selected, along with a basic description of their fragrance and the scoring method for each tester, are shown in

Table A12. Olfactory materials and marking schedule for odor recognition.

Material	Characteristic Odor 3 Points	2 Points	1 Point
Citral	Lemon	Orange	Citrus, fruity
Ethyl caproate	Apple	Pear	Fruity
Fresh orange juice	Orange	Lemon	Citrus, fruity
Caramel syrup	Caramel	Vanilla, malt, toasted	Sweet
Butyric acid	Rancid butter	Parmesan cheese	Vomit, baby burp
Acetic acid	Acetic acid	Vinegar	Sour/off wine
cis-3-hexen-1-ol	Green grass	Green beans	Green vegetable
TCA	Musty, Moldy	Cork taint	Chlorine
Vanillin	Vanilla	Custard, dessert	Sweets, candy

.

Evaluation of Basic Training

In the tests described in Sections “Basic Taste Recognition”, “Ranking Tests”, and “Odor Recognition” in Section 2.5.5, the maximum number of points that a candidate could collect is 53. A tester’s training is considered successful if the score they collect is higher than 65% (more than 35 points). The ranking of the candidates’ scores was taken into account in their final selection.

2.5.6. Training and Identification of Wine Defects Using Aqueous Solutions

Aqueous solutions were created to train tasters on wine defects. The solutions were presented to the candidate assessors, and the instructor explained each aroma. For each defect, three solutions were prepared at low, medium, and high intensities. Testers were then asked to rank the samples within each triad based on increasing intensity. The substances used and the concentrations corresponding to low, medium, and high intensities are shown in

Table A13. The substances used and the concentrations corresponding to low, medium, and high intensities for the training and identification of wine defects using aqueous solutions.

Attributes	Descriptor Examples	Material	Low Intensity	Medium Intensity	High Intensity
Sulfur dioxide	Sulfur dioxide	Sulfur dioxide	0.1 g/L	0.2 g/L	0.3 g/L
Acetic	Acescent (glue, nail, polish, solvent)	Ethyl acetate	0.25 mg/L	0.50 mg/L	0.75 mg/L
Moldy/Earthy	Dusty, musty, damp cellar	Trichloroanisole (TCA)	5 ng/L	15 ng/L	25 ng/L
Oxidized	Stale, rancid, overripe apple	Ethanal	0.1 g/L	0.2 g/L	0.3 g/L
Animal	Stale, rancid, overripe apple	Mix 4-ethylphenol/4-ethylguajacol	0.5 mg/L	1.0 mg/L	1.5 mg/L

.

2.5.7. Identification of Wine Defects and Wine Characteristics Using Spiked Wine

Five different wine defects and three additional wine characteristics were selected, consisting of samples with varying fruit intensities, different tannin intensities, and whether they were aged in barrels or not. A triangle test was performed for each group of samples. These triangle tests were conducted in duplicate to determine whether the candidate assessors could replicate their results consistently. The sample sets and the method of presentation were based on international standards [38], with necessary adaptations from the OIV recommendations [31], as shown in

Table A14. The substances used and the concentrations for the identification of wine defects and wine characteristics using spiked wine.

a/a	Group Characterization	Batch 1	Batch 2	Different Sample out of the Three
1	Fresh wine—Wine with barrel	Fresh wine	Fresh Wine spiked with 500 mg/L Vanillin	Batch 2
2	Oxidized wine	Fresh wine	Wine spiked with 0.5 g/L Ethanal	Batch 2
3	Different intensity of fruity	Fresh wine	Wine diluted with water (80% v/v)	Batch 2
4	Animal defect	Fresh wine	Wine spiked with 700 mg/L 4-ethylphenol and 0.070 mg/L 4-ethylguajacol	Batch 2
5	Acetic	Fresh wine	Wine spiked with 1.2 g/L acetic acid	Batch 2
6	Different tannins	Fresh red wine	Wine spiked with 1.5 g/L oak tannin	Batch 2
7	Sulfur dioxide	Fresh wine	Wine spiked with 200 mg/L potassium disulfite	Batch 2
8	TCA	Fresh wine	Wine spiked with 15 ng/L TCA	Batch 2

.

The candidate assessors were rated according to the following scale: each tester received 6 points if they provided a correct response for both sample set #1 and sample set #2, 4 points if they had a correct response for either sample set #1 or sample set #2, 0 points if they had an incorrect response for both sample set #1 and sample set #2, and 0.5 points if they accurately described the difference.

2.5.8. Sensory Evaluation of Different Types of Tannins

The candidates were asked to taste solutions of anthocyanidic tannins (from grape skin), gallic tannins, and ellagic tannins (from oak) at a concentration of 0.1 g/L. They were instructed to evaluate these solutions and write down all the descriptive words that could characterize the samples. The training leader then judged whether the descriptions were correct or incorrect. The responses of the testers were rated on a scale from 0 to 2 points for each type of tannin, resulting in a maximum score of 6.

2.5.9. Ranking Test of Wines with Different Intensities of Oak Barrel Aroma

The candidate assessors were asked to rank four wine solutions appropriately spiked to create solutions of varying oak barrel aroma intensities. This was achieved by spiking a wine sample with a mixture of vanillin and whiskey lactone (50–50% w/v). The concentrations of these substances in the final sample are detailed in Table A11. Detailed information about the compounds is shown in Table A1. Each tester’s score ranged from 0 to 4 points, as specified in Table A7.

2.5.10. Marking the Schedule Summary

The maximum score of all the above tests when examining the testers is shown in

Table A15. Maximum test scores for the evaluation of training and the selection of assessors.

Test No.	Paragraph	Title	Maximum Score Possible
1	Section 2.5.5. “Odor Recognition”	Odor recognition	27
2	Section 2.5.5. “Basic Taste Recognition”	Basic taste recognition	6
3	Section 2.5.5. “Ranking Tests”	Ranking of basic taste	16
4	Section 2.5.9.	Ranking of barrel flavor	4
5	Section 2.5.7.	Triangle test—flavor	48
6	Section 2.5.7.	Triangle test—Description	4
7	Section 2.5.8.	Descriptive different tannins	6

. Guidelines for the selection of panel members, according to the marking schedule, are listed in

Table A16. Marking schedule for assessor selection.

Test No.	Title	Maximum Score Possible	Recommended Pass Mark	Pass Score
1 + 2 + 3 + 4	Recognition and ranking tests	53	65%	≥35/53
7 + 8 + 9	Descriptive tests	10	50%	≥5/10
7 + 8	Triangle test—Wine	48	65%	≥32/48

and are as follows:

(a)

An overall score of 65% or greater in a combination of odor recognition, basic taste recognition, ranking of basic taste, and ranking of barrel flavor.

(b)

An overall score of 50% or greater for the descriptive tests.

(c)

If the panel members obtained correct scores in at least 65% of the triangle tests.

(d)

Those candidates who achieve less than 65% of the total score are considered unsuitable as sensory evaluators and require additional training [38].

2.5.11. Assessors’ Selection—Creation of New Panels

The results of the candidates’ scores were evaluated to create two new sensory panels. The candidate assessors were ranked in descending order of score, and subsequently, two new panels were formed: Panel C comprised the 10 highest-scoring assessors, while Panel D consisted of the 10 lowest-scoring assessors.

2.6. Wine’s Sensory Analysis Method

2.6.1. General Instructions

The instructions provided to the assessors and the general rules followed are outlined below.

The optimal time for conducting a sensory analysis is between 10:00 and 12:00 in the morning. In order to ensure the best possible performance, assessors should refrain from wearing perfumes, be in good physical and psychological condition, and avoid smoking for at least half an hour before the test.

The sensory analyses take place in a dedicated test room designed according to the specifications outlined in the relevant standard documents [43,44]. The testing environment is well ventilated and air-conditioned, maintaining a temperature range of 20–25 °C. During the tests, the samples are presented at a consistent temperature of 15 °C, each labeled with a random three-digit number. ISO 3591 tasting glasses, sealed with watch glasses, are used for the analysis. Mineral water serves as a palate cleanser, and each session accommodates a maximum of four samples, with a limit of three sessions per day.

2.6.2. Sensory Analysis Methodology

The wine sensory analysis method developed includes 3 negative attributes and 11 positive attributes. The three negative qualities are appearance defect, aroma defect, and taste defect. The appearance defects include a hazy appearance, browning, and a pinking effect. Aroma disadvantage and taste disadvantage encompass characteristics such as moldy/earthy, acetic, reduced, oxidized, animal, lactic, and TCA. On the other hand, the 11 positive qualities consist of turbidity, color, hue, fruity aroma, barrel aroma, flavor intensity, sour, sweet, bitter, astringency, and aftertaste duration.

In order to assess these attributes, a tester sheet was employed for all attributes, where each attribute was linked to a continuous scale of 0–10. The assessors were instructed to mark an “x” or a vertical line on the corresponding scale position to indicate the perceived intensity of the attribute. The profile sheet is provided in the appendix (Figure A1).

The sensory analysis process begins with a prolonged visual observation, focusing on parameters such as appearance defect, turbidity, color, and hue. Subsequently, the assessor evaluates the olfactory parameters (including aroma defect, fruity aroma, and barrel aroma) by inhaling without swirling the glass and then inhaling after swirling it. Finally, after taking the first mouthful, the wine is allowed to warm briefly while being moved around the mouth. The assessor then inhales through the nose and spits. The taste attributes (including taste defect, flavor intensity, sour, sweet, bitter, astringency, and aftertaste duration) are recorded. When assessing negative attributes, the assessor notes the intensity of the most significant defect (if multiple exist) and selects the corresponding defect characterization from the profile sheet. If the defect is not described on the sheet, the term “other” is chosen.

2.6.3. Use of the Data by Panel Leader

The panel leader measures the intensities of each property to one decimal place and enters the data into a computer program. For each sample, the median value (Me) and the robust coefficient of variation (CVR%) are calculated. Specifically, the robust coefficient of variation for a potential defect, a fruity aroma, and sourness must not exceed 20.0%. If it does, the panel leader must repeat the evaluation of the specific sample in another tasting session [32].

In this study, data entry into the computer program utilized specialized software for the organoleptic analysis (Fizz Calculations, Biosystemes, Couternon, France). The calculations for the median value and robust coefficient of variation were performed using the calculation program of the International Olive Oil Council (IOC).

2.6.4. Assessors’ Training on the Profile Sheet

The panel leader proceeded to train the two test groups, C and D, in the use of the profile sheet. Four sessions were conducted, involving the analysis of wine samples and references. In order to prepare the assessors for using the continuous scale, training instructions from the organoleptic evaluation of olive oil [34] were employed for the aromatic attributes. Additionally, recent studies [18] were utilized to train the testers on the continuous scale for taste parameters (sour, sweet, and astringent), and concentrations for bitter were also incorporated, as indicated in

Table A17. The substances used and the concentrations for training with the continuous scale.

		Concentration in Water at Room Temperature g/L
Taste	Material	Level 1	Level 2	Level 3	Level 4	Level 5
Sweet	D(+)-Glucose	15.6	46,9	78.1	109.3	156.3
Sour	Tartaric acid	4.0	5.0	6.0	7.0	8.0
Astringent	Grape tannin	0.5	2.2	3.6	5.0	6.0
Bitter	Caffeine	0.05	0.30	0.50	0.70	1.00

.

2.7. Validation of the Method

For method validation and to check the performance of each assessor and each team, we selected characteristic defects (such as “animal” defects, with the use of 4-ethylphenol), an olfactory parameter (fruity), and a taste parameter (sour). Seven wines derived from Greek grape varieties were utilized for the calculations. The process unfolded as follows: Each day, a set of four samples was provided to the tasters. Among these, two were identical (duplicate), one sample was repeated daily (to evaluate reproducibility), and the fourth sample was randomly chosen. Each sample was assigned a unique three-digit code, and the presentation followed the principles of the Latin square design.

2.7.1. Checking the Performance of Each Taster

In order to assess the performance of each tester, we employed the replicate analysis method. In this approach, a sample intended for analysis is presented to the testers twice during the same session, each time with distinct three-digit codes. The taster’s performance is influenced by two key factors:

Deviation from itself: This refers to any variation observed when analyzing the same sample at two different time points.

Deviation from the group (the panel): This accounts for differences in performance compared to the overall panel during the same session.

In order to measure these differences, the precision number (PN) and the deviation number (DN) were used.

Checking the Taster’s Precision

The precision (repeatability and in-laboratory intermediate precision) of all tasters was determined using the replicate analysis method. Repeatability for each taster was assessed by comparing the intensity scores when analyzing duplicate samples.

Intermediate precision was evaluated over time using the precision number (PN), which considers the intensities assigned by the taster to a set of six duplicate samples on different days. The PN is defined as

$$PN={\displaystyle \frac{{\sum }_{i=1}^n{(x_{i,1}-x_{i,2})}^2}{n}}$$

where x_i,1 and x_i,2 are the values given by the taster to the first and second assessment of a duplicated sample, and n is the number of differences (x_i,1 − x_i,2), corresponding to the number of duplicated samples analyzed.

The criterion of acceptance of the taster was PNt ≤ 2.0.

Checking the Taster’s Trueness

Trueness refers to the agreement between the average value of a large series of measurements and the accepted reference value or the “true value”. In our specific case, where no reference materials or interlaboratory proficiency testing data were available, we considered the panel median for the specific attribute as the true value.

In order to estimate trueness, the deviation number (DN) was calculated based on the analysis of six duplicate samples on different days.

The DN is defined as

$$DNt={\displaystyle \frac{{\sum }_{i=1}^n{(x_i-{Me}_i)}^2}{n}}$$

where DNt is the deviation number of a taster for the specific attribute in the first replicate of the duplicate sample “I”, x_i is the intensity score given by the taster for the specific attribute, Mei is the value of the panel median for the specific attribute, and n is the number of differences considered for the calculation (n = 6 in this calculation).

The criterion of acceptance of the taster was DNt ≤ 2.0.

Checking the Panel’s Precision

The repeatability of the panel was evaluated by comparing the pair of medians obtained by analyzing a sample in duplicate with the calculation of the normalized error (En).

The precision of the panel was checked by comparing the pairs of medians obtained by analyzing six duplicate samples (12 samples analyzed in total) with the calculation of the precision number for the panel (PNp).

More specifically, the En was calculated according to the following formula:

$${En}_p={\displaystyle \frac{\left(|{Me}_1-{Me}_2|\right)}{\sqrt{{U_1^2+U}_2^2}}}$$

where En is the normalized error of the panel for the specific attribute; Me₁ and Me₂ are the medians obtained by the panel for the specific attribute in the first and second assessment of a sample, respectively; U₁ and U₂ are the respective expanded uncertainties calculated as c × s₁ and c × s₂, with c = 1.96 for a 95% probability; and s₁ and s₂ are the experimental robust standard deviation values of the medians Me₁ and Me₂, respectively, for the attribute.

The criterion of the acceptance of the panel was En ≤ 1.0.

The precision number for the panel (PNp) was calculated according to the following formula:

$$PNp={\displaystyle \frac{{\sum }_{i=1}^n{({Me}_{i,1}-{Me}_{i,2})}^2}{n}}$$

where PNp is the indicator of the consistency of a panel “p” for a specific attribute in evaluating a sample in duplicate; Me₁ and Me₂ are the medians of the panel for the specific attribute in the first and second assessment of the duplicated sample, respectively; and n is the number of duplicated samples taken into account (in this case, six duplicate samples, n = 6).

The criterion of acceptance was PNp ≤ 2.0.

Calculation of Reproducibility

In order to evaluate the reproducibility of the sensory analysis method, six analyses were performed on different days using the same red dry wine from the Agiorgitiko variety, which had matured in oak barrels. Based on the results obtained by the panel, the median, standard deviation, and relative standard deviation were calculated.

2.8. Sensory Analysis of Nemea Wines—Statistical Analysis

Panel C applied the developed method to sensorially analyze the 25 commercial wines from AOC Nemea. Each sample was analyzed in duplicate, and the median value was calculated for each organoleptic attribute. Subsequently, the average of the two values for each sensory attribute was calculated.

Additionally, the data underwent a one-way analysis of variance (ANOVA) for the products to assess the discriminatory ability of each attribute [25]. Finally, principal component analysis (PCA) was performed using JMP software 18.0.1. version(SAS, Campus Drive Cary, NC, USA).

3. Results

3.1. Taster’s Performance during Basic Training

For the ageusia tests, all tasters in Team A had a pass rate of 80 to 100%. The average successful responses of Team A were 90%. The Team B tasters had a success rate of 60% to 90%. More specifically, two candidate assessors had a percentage of less than 80%. The average of the successful responses of Team B was 78%. For the two candidate assessors who had a percentage of less than 80%, a repetition of the test was carried out, in which they had a percentage of successful answers greater than or equal to 80%.

For the anosmia tests, the Team A tasters had a success rate of 63 to 100%. The average of the successful responses of Team A was 93%. The Team B tasters had a success rate of 38 to 100%. More specifically, two candidate assessors from Team A and six from Team B had a percentage of less than 80%. The average of the successful responses of Team B was 75%. The candidate assessors who had a percentage of less than 80% were retested and given a short training session by the team leader, in which they had a percentage of successful responses greater than or equal to 80%.

In the Ishihara test, the two teams had exactly the same results. All candidates had 100% success results, except two—one from Team A and one from Team B—who had 94% success.

Team A had a lower perception threshold than Team B in all five basic tastes, as shown in

Table 1. The perception threshold of each team in the five basic tastes.

		Team A	Team B
Taste	Material	Threshold g/L	Threshold g/L
Sour	Citric acid	0.12	0.15
Bitter	Caffeine	0.061	0.15
Salty	Sodium chloride	0.034	0.073
Sweet	Sucrose	0.28	0.30
Umami	Monosodium L-glutamate	0.081	0.13

.

In the tests that determined organoleptic awareness, during the process of detecting taste stimuli using triangle tests, all but one taster in Team A achieved success, resulting in a total of 59 correct answers out of 60. Similarly, all tasters in Team B were successful, except for two tasters, yielding a total of 55 correct answers out of 60.

In the taste identification tests, each taster had to identify 10 flavored solutions. For each correct identification, the taster received 1 point, resulting in a maximum score of 10 points per taster. Team A scored 87 points, and Team B scored 78 points.

In the ranking tests, all tasters successfully identified the color of red wines, leading to identical performances for Teams A and B. For odor discrimination, which involved four aromatic substances (acetic acid, TCA, ethyl acetate, and ethyl phenol), each taster could earn up to 16 points. Team A had four tasters with perfect scores (16/16) and an average score of 14.6. In contrast, Team B had no tasters with a perfect pass and an average score of 12.4.

For taste and astringency discrimination, which included the five basic tastes and astringency, each candidate assessor had the opportunity to earn 24 points. Team A had four tasters with perfect scores (24/24) and an average score of 22.8, while Team B also had four tasters with a perfect pass (24/24) and an average score of 22.1.

Finally, in the odor description tests, each candidate assessor could earn up to 30 points (10 aromas with a maximum score of 3 points per aroma). The scores of Team A ranged from 12 to 30 points, with an average of 19.2, while those for Team B ranged from 9 to 19 points, averaging 14.0.

3.2. Evaluation of Training

In the evaluation of basic training (“recognition and ranking tests”), there were 53 available points. Successful training was considered when the percentage was greater than or equal to 65%, equivalent to 35 points. All the tasters in Team A achieved successful results. The minimum score observed was 35 (65%), and the maximum score reached 53 (100%), resulting in an average of 44.8 points (84%). In contrast, Team B had five tasters with successful outcomes, while the remaining five tasters did not perform well. The minimum score for Team B was 24 (45%), the maximum was 38 (72%), and the average score stood at 33.5 points (63%). These results are presented in

Table 2. Taster scores for each stage of basic training, total score, average score for both teams, evaluation of each training stage, and overall taster evaluation.

	Score Evaluation				Scores
Taster’s Code	Total Evaluation	Recognition and Ranking Tests	Triangle Test—Wine	Descriptive Tests	Total Score	Recognition and Ranking Tests	Triangle Test—Wine	Descriptive Tests
Team A
CA01	Pass	Pass	Pass	Pass	110	52	48	10
CA02	Pass	Pass	Pass	Pass	94	42	42	10
CA03	Pass	Pass	Pass	Pass	103	45	48	10
CA04	Pass	Pass	Pass	Pass	109	53	48	8
CA05	Pass	Pass	Pass	Pass	100	44	46	10
CA06	Pass	Pass	Pass	Pass	108	53	46	9
CA07	Pass	Pass	Pass	Pass	100	46	46	8
CA08	Pass	Pass	Pass	Pass	77	40	32	5
CA09	Pass	Pass	Pass	Pass	84	38	36	10
CA10	Pass	Pass	Pass	Pass	79	35	38	6
Average					96.3	44.8	43.0	8.5
Team B
CB01	Fail	Fail	Pass	Pass	76	30	40	6
CB02	Pass	Pass	Pass	Pass	88	38	40	10
CB03	Fail	Pass	Fail	Pass	70	35	28	7
CB04	Pass	Pass	Pass	Pass	83	38	36	9
CB05	Fail	Fail	Pass	Pass	80	34	40	6
CB06	Fail	Fail	Pass	Pass	72	24	42	6
CB07	Pass	Pass	Pass	Pass	81	36	38	7
CB08	Fail	Fail	Fail	Fail	64	33	28	3
CB09	Fail	Fail	Pass	Pass	83	32	46	5
CB10	Fail	Pass	Fail	Fail	66	35	26	5
Average					76.0	33.5	36.4	6.1

.

When identifying wine defects and wine characteristics using spiked wine, the number of available points for each candidate assessor was 48, with a minimum score for successful results set at 32 points. All tasters in Team A achieved successful results, with an average score of 43.0 points. The lowest individual taster score in Team A was 32 points, while the highest reached 48 points. In Team B, seven tasters achieved successful results, while the remaining three tasters did not perform well. The mean score for Team B was 36.4 points, with a minimum individual candidate assessor score of 26 points and a maximum of 46 points, as shown in Table 2.

In the descriptive tests, which focused on the accurate identification of positive or negative wine aromas, as well as the description of different tannins, the maximum achievable score was 10 points. Results equal to or greater than 5 points were considered successful. In Team A, all tasters achieved successful results, with an average score of 8.5 points. In Team B, eight tasters succeeded, while two tasters did not, resulting in an average score of 6.1 points, as shown in Table 2.

In the total evaluation of each candidate assessor’s training, the training was considered successful if the candidate had successful results in all three areas: the recognition and ranking tests, the triangle test for wine, and the descriptive tests. According to this criterion, all 10 candidate testers from Group A achieved successful results, while only 3 candidate testers from Group B met the criteria. Subsequently, a ranking of the total scores was conducted for the 13 successful candidate testers to facilitate the final selection process. The top 10 assessors formed Panel C, comprising 8 tasters from Group A and 2 from Group B. The remaining tasters constituted Panel D, which served the research purposes of this study.

3.3. Validation of the Method

3.3.1. Checking the Panel’s Performance

Panel C yielded successful results in all duplicate sample measurements, both for normalized error and precision number, across all three organoleptic parameters tested. However, Panel D exhibited unsuccessful precision number results in three duplicate measurements, specifically for the defect (animal) parameter. Notably, all other duplicate measurements in Panel D were successful, demonstrating consistent performance in terms of both normalized error and precision number for the three organoleptic parameters, as detailed in

Table 3. Normalized error and precision values for each panel in each duplicate analysis of the three organoleptic parameters.

Attribute	Fruity		Defect (Animal)		Sour
Samples Number	Normalized Error Fruity (En ≤ 1.0)	Precision Number Fruity (≤2.00)	Normalized Error Defect (En ≤ 1.0)	Precision Number Defect (≤2.00)	Normalized Error Sour (En ≤ 1.0)	Precision Number Sour (≤2.00)
Panel C
REP_1-2	0.95	0.41	0.00	0.00	0.38	0.16
REP_3-4	0.30	0.25	0.96	1.00	0.28	0.16
REP_5-6	0.36	0.16	0.27	0.25	0.26	0.04
REP_7-8	0.08	0.01	0.00	0.00	0.00	0.00
REP_9-10	0.27	0.09	0.00	0.00	0.38	0.04
REP_11-12	0.45	0.16	0.00	0.00	0.23	0.01
Panel D
REP_1-2	0.40	0.64	0.74	2.56	0.04	0.01
REP_3-4	0.50	0.25	0.69	2.89	0.16	0.09
REP_5-6	0.27	0.16	0.33	0.81	0.19	0.16
REP_7-8	0.23	0.16	0.25	0.64	0.41	0.49
REP_9-10	0.03	0.01	0.65	5.29	0.37	1.00
REP_11-12	0.23	0.09	0.53	1.69	0.55	1.96

.

The overall performance of the panel was assessed using the precision number (PNp). Panel C achieved successful results (PNp ≤ 2.00) across all three organoleptic parameters. In contrast, Panel D exhibited an unsuccessful outcome in defect determination but achieved successful results for the other two organoleptic parameters. Notably, the precision number for Panel C was lower than that of Panel D across all three organoleptic parameters, as shown in

Table 4. Panel performance and taster performance for the three sensory attributes.

Attribute	Fruity		Defect (Animal)		Sour
	Precision Number	Deviation Number	Precision Number	Deviation Number	Precision Number	Deviation Number
Panel Performance
Panel C	0.18		0.21		0.07
Panel D	0.22		2.31		0.62
Taster Performance
Panel C
CC01	1.09	0.09	0.73	0.44	0.28	0.51
CC02	0.49	0.50	0.33	1.29	0.65	0.24
CC03	0.50	0.26	0.85	0.52	0.33	0.23
CC04	0.53	0.27	1.16	1.58	0.17	0.85
CC05	0.25	1.44	1.99	1.26	0.39	0.21
CC06	2.00	1.74	0.92	0.80	0.71	0.72
CC07	0.83	1.45	1.31	0.57	0.43	0.11
CC08	0.24	1.13	0.02	0.63	0.34	1.26
CC09	0.33	1.88	0.41	0.47	1.79	0.89
CC10	0.31	1.00	0.19	0.10	0.25	0.53
Panel D
CD01	2.06	3.62	1.83	3.87	0.59	1.84
CD02	6.82	2.53	3.03	11.89	1.82	7.06
CD03	5.89	4.64	0.34	8.27	0.68	1.08
CD04	7.52	2.36	3.43	2.09	0.29	5.05
CD05	0.54	1.20	27.67	3.46	2.64	1.15
CD06	2.70	3.03	0.56	0.20	3.67	1.66
CD07	14.09	6.72	0.20	6.76	0.38	1.57
CD08	1.88	2.60	4.51	4.17	1.34	1.84
CD09	18.33	2.95	16.81	1.44	15.28	3.33
CD10	0.98	0.32	4.97	0.45	0.89	1.41

.

3.3.2. Checking the Taster’s Performance

Among the 10 tasters in Panel C, successful results were observed for all three organoleptic parameters. Specifically, the precision number (PN) remained consistently at or below 2.00, and the deviation number (DN) also met the same criterion, as shown in Table 4. In contrast, none of the 10 tasters in Panel D achieved successful results across all three organoleptic parameters. For the fruity attribute, seven tasters failed to meet the precision number threshold, and eight tasters fell short in terms of the deviation number. Regarding the animal defect, six tasters did not achieve the required precision number, while seven tasters exceeded the acceptable deviation number. Finally, for the sour attribute, three tasters failed to meet the precision number requirement, and an equal number of tasters fell short in terms of the deviation number, as shown in Table 4.

3.3.3. Reproducibility

Panel C exhibited a relative standard deviation (RSD) significantly below 10%, spanning a range from 3.6% to 5.8% across the three organoleptic parameters tested. In contrast, Panel D displayed a RSD of approximately 20%, with values ranging from 18.4% to 24.5% for the same set of organoleptic parameters, as indicated in

Table 5. Descriptive statistics for each panel and sensory attribute.

	Defect Aroma (Median)	Fruity Aroma (Median)	Sour (Median)
Panel A
Rep-1	2.2	5.1	4.8
Rep-2	2.4	5.3	4.9
Rep-3	2.3	5.4	4.9
Rep-4	2.6	4.9	5.2
Rep-5	2.3	5.1	4.7
Rep-6	2.4	5	5.1
Mean	2.37	5.13	4.93
Standard Deviation	0.14	0.19	0.19
Relative Standard Deviation	5.8%	3.6%	3.8%
Panel B
Rep-1	4.0	5.5	4.5
Rep-2	4.7	5.7	4.7
Rep-3	5.4	5.5	5.1
Rep-4	4.5	4.9	4.8
Rep-5	3	3.1	3
Rep-6	4.4	3.4	3.5
Mean	4.33	4.68	4.27
Standard Deviation	0.8	1.15	0.83
Relative Standard Deviation	18.4%	24.5%	19.4%

.

3.3.4. Sensory Analysis of Wines

A total of 25 commercial Nemea wines were sensorially analyzed by Panel C. The results of the median values for each organoleptic parameter, as well as the retail price, are shown in

Table 6. Sensory analysis results and retail prices of Nemea wines.

	Retail Price (EUR)	Defect Appearance	Defect Aroma	Defect Taste	Turbidity	Color	Hue	Fruity Aroma	Barrel Aroma	Flavor Intensity	Sour	Sweet	Bitter	Astringent	Aftertaste
KK001	8.9	0.0	0.0	0.00	8.1 ± 0.01	7.2 ± 0.05	6.0 ± 0.01	5.0 ± 0.06	1.8 ± 0.12	4.2 ± 0.00	4.0 ± 0.06	0.9 ± 0.04	0.8 ± 0.08	3.1 ± 0.15	4.6 ± 0.13
KK002	10.5	0.0	1.6 ± 0.01	0.00	8.0 ± 0.21	7.1 ± 0.16	5.8 ± 0.12	5.2 ± 0.05	3.4 ± 0.06	4.3 ± 0.05	4.1 ± 0.00	0.6 ± 0.01	1.1 ± 0.05	3.6 ± 0.01	5.1 ± 0.09
KK003	8.1	0.0	0.0	0.00	8.2 ± 0.02	5.4 ± 0.05	3.6 ± 0.04	4.2 ± 0.11	3.3 ± 0.04	4.4 ± 0.10	4.4 ± 0.04	0.9 ± 0.10	1.9 ± 0.05	2.9 ± 0.02	4.5 ± 0.01
KK004	8.9	0.0	3.2 ± 0.13	0.00	8.1 ± 0.13	7.2 ± 0.04	6.0 ± 0.04	3.3 ± 0.03	1.4 ± 0.01	2.6 ± 0.02	3.6 ± 0.15	0.9 ± 0.03	1.1 ± 0.00	3.4 ± 0.14	4.4 ± 0.04
KK005	12.1	0.0	0.0	0.00	8.1 ± 0.14	8.1 ± 0.10	7.3 ± 0.09	3.2 ± 0.02	4.6 ± 0.01	4.0 ± 0.10	3.8 ± 0.07	1.6 ± 0.06	1.1 ± 0.00	4.1 ± 0.02	5.0 ± 0.03
KK006	8.1	0.0	0.0	0.00	8.1 ± 0.11	7.0 ± 0.07	6.4 ± 0.18	3.7 ± 0.06	4.9 ± 0.02	4.0 ± 0.02	4.0 ± 0.07	0.9 ± 0.02	0.9 ± 0.01	3.8 ± 0.03	4.6 ± 0.06
KK007	7.3	0.0	0.0	0.00	8.0 ± 0.04	6.6 ± 0.06	4.2 ± 0.08	4.8 ± 0.05	3.9 ± 0.04	4.1 ± 0.13	4.1 ± 0.08	1.1 ± 0.09	0.9 ± 0.00	3.3 ± 0.11	4.6 ± 0.06
KK008	12.1	0.0	0.0	0.00	8.4 ± 0.01	8.2 ± 0.10	7.6 ± 0.02	6.0 ± 0.02	2.6 ± 0.03	4.1 ± 0.04	4.4 ± 0.18	0.8 ± 0.07	0.9 ± 0.08	3.3 ± 0.16	4.8 ± 0.03
KK009	18.6	0.0	0.0	0.0	7.9 ± 0.03	6.8 ± 0.09	3.2 ± 0.08	4.6 ± 0.09	2.6 ± 0.15	3.7 ± 0.05	3.6 ± 0.05	0.7 ± 0.06	0.9 ± 0.01	2.9 ± 0.09	4.0 ± 0.01
KK010	5.7	0.0	0.0	0.0	8.0 ± 0.03	4.8 ± 0.02	3.4 ± 0.08	4.1 ± 0.00	2.0 ± 0.03	3.2 ± 0.14	3.8 ± 0.11	0.5 ± 0.11	1.2 ± 0.04	3.2 ± 0.16	4.3 ± 0.11
KK011	8.9	0.0	0.0	0.0	8.4 ± 0.07	6.5 ± 0.01	4.9 ± 0.11	5.5 ± 0.08	2.4 ± 0.04	4.2 ± 0.10	4.1 ± 0.06	0.4 ± 0.02	1.2 ± 0.06	3.6 ± 0.01	4.7 ± 0.06
KK012	17.7	0.0	0.0	0.0	8.7 ± 0.07	6.7 ± 0.13	5.5 ± 0.09	3.6 ± 0.04	4.8 ± 0.01	4.3 ± 0.08	3.9 ± 0.10	0.9 ± 0.07	0.9 ± 0.1	3.1 ± 0.11	4.8 ± 0.04
KK013	12.9	0.0	0.0	0.0	8.5 ± 0.06	6.5 ± 0.05	4.1 ± 0.02	4.4 ± 0.02	4.8 ± 0.08	3.9 ± 0.18	4.5 ± 0.09	0.8 ± 0.07	1.0 ± 0.12	3.2 ± 0.11	4.7 ± 0.03
KK014	9.7	0.0	0.0	0.0	8.4 ± 0.13	7.3 ± 0.01	4.9 ± 0.03	3.8 ± 0.05	4 ± 0.06	3.6 ± 0.08	3.9 ± 0.02	0.7 ± 0.06	1.1 ± 0.07	3.7 ± 0.09	4.8 ± 0.04
KK015	10.5	0.0	0.0	0.0	8.1 ± 0.08	6.4 ± 0.06	1.8 ± 0.11	4.8 ± 0.07	3.2 ± 0.07	3.6 ± 0.09	4.0 ± 0.13	1.0 ± 0.05	0.8 ± 0.04	3.2 ± 0.05	4.4 ± 0.01
KK016	12.1	0.0	0.0	0.0	8.3 ± 0.06	7.3 ± 0.05	5.9 ± 0.09	5.5 ± 0.05	3.4 ± 0.03	4.1 ± 0.00	4.3 ± 0.11	0.7 ± 0.15	1.0 ± 0.09	3.6 ± 0.04	4.6 ± 0.03
KK017	7.3	0.0	1.5 ± 0.00	0.0	8.7 ± 0.04	3.9 ± 0.06	5.0 ± 0.02	3.7 ± 0.03	0.0 ± 0.00	3.3 ± 0.08	3.7 ± 0.02	0.6 ± 0.03	1.0 ± 0.00	2.8 ± 0.03	3.7 ± 0.01
KK018	11.3	0.0	0.0	0.0	8.2 ± 0.01	7.1 ± 0.04	6.0 ± 0.13	4.3 ± 0.02	3.2 ± 0.03	4.3 ± 0.09	3.7 ± 0.12	2.0 ± 0.03	0.6 ± 0.06	2.8 ± 0.11	4.0 ± 0.03
KK019	8.9	0.0	0.0	0.0	8.2 ± 0.17	7.4 ± 0.08	6.1 ± 0.05	4.7 ± 0.12	3.3 ± 0.03	4.1 ± 0.15	4.3 ± 0.11	0.5 ± 0.02	0.7 ± 0.07	3.4 ± 0.04	4.9 ± 0.16
KK020	7.3	0.0	0.0	0.0	8.3 ± 0.17	6.8 ± 0.06	5.6 ± 0.00	5.4 ± 0.01	3.3 ± 0.14	4.8 ± 0.09	4.0 ± 0.11	1.3 ± 0.05	0.6 ± 0.11	2.9 ± 0.03	4.4 ± 0.03
KK021	28.2	0.0	1.6 ± 0.01	0.00	7.9 ± 0.08	6.0 ± 0.09	2.4 ± 0.06	3.3 ± 0.05	3.7 ± 0.06	3.5 ± 0.10	3.7 ± 0.16	1.0 ± 0.05	0.5 ± 0.01	3.1 ± 0.09	4.1 ± 0.05
KK022	28.2	0.0	1.1 ± 0.03	0.00	8.3 ± 0.11	6.4 ± 0.06	2.5 ± 0.06	3.8 ± 0.01	4.5 ± 0.04	3.7 ± 0.11	4.0 ± 0.07	0.9 ± 0.00	0.7 ± 0.02	3.1 ± 0.04	4.4 ± 0.06
KK023	8.9	0.0	0.0	0.00	8.0 ± 0.02	8.3 ± 0.12	6.9 ± 0.03	4.6 ± 0.05	4.0 ± 0.10	3.7 ± 0.05	4.3 ± 0.15	1.1 ± 0.03	0.8 ± 0.09	3.9 ± 0.10	5.1 ± 0.02
KK024	14.5	0.0	0.0	0.00	8.1 ± 0.18	6.8 ± 0.12	5.3 ± 0.12	4.9 ± 0.07	3.6 ± 0.03	4.2 ± 0.05	5.0 ± 0.1	0.6 ± 0.00	0.8 ± 0.02	3.2 ± 0.03	4.6 ± 0.06
KK025	8.1	0.0	0.0	0.00	8.1 ± 0.21	6.8 ± 0.01	5.8 ± 0.04	5.0 ± 0.12	2.6 ± 0.03	4.1 ± 0.01	4.0 ± 0.02	1.6 ± 0.07	0.7 ± 0.03	3.5 ± 0.02	5.1 ± 0.05

.

Five wines exhibited organoleptic aroma defects at a low intensity. The range of organoleptic defects spanned from 1.1 to 3.2. The tasters characterized two defects as animal, two as oxidized, and one as reductive.

The two samples displaying an animal defect underwent analysis using GC-MS/MS to determine the aroma compounds. The detailed analysis methodology, including the extraction method, GC-MS/MS operating conditions, and selected mass transitions for each compound, was described in a previous study by the authors [45]. In the first sample, with a sensory aroma defect intensity of 3.2, the following concentrations were observed: 4-ethylphenol (223 μg/L), 4-ethylguajacol (90 μg/L), and 4-vinylphenol (1119 μg/L). In the second sample, with a sensory aroma defect intensity of 1.1, the concentrations were as follows: 4-ethylphenol (127 μg/L), 4-ethylguajacol (31 μg/L), and 4-vinylphenol (314 μg/L).

Regarding the appearance organoleptic parameters, both color and hue exhibited substantial variations. Specifically, color ranged from 3.9 to 8.3, with an average value of 6.8, while hue ranged from 1.8 to 7.6, with an average value of 5.0.

For the aroma attributes, fruity aroma spanned between 3.2 and 6.0, with an average value of 4.5, and barrel aroma ranged from 0.0 to 4.9, with an average value of 3.3. Notably, seven samples showed higher barrel aroma intensity than fruity aroma.

The flavor intensity varied between 2.6 and 4.8, with an average value of 3.9. In terms of taste attributes, sour varied from 3.6 to 5.0, with an average value of 4.1; sweet ranged from 0.4 to 2.0, with an average value of 0.9; bitter exhibited similar values to sweet, ranging from 0.5 to 1.9, with an average value of 0.9; astringent values ranged between 2.8 and 4.1, with a mean value of 3.3; and aftertaste spanned from 3.7 to 5.1, with a mean value of 4.6.

3.3.5. Data Analysis

A one-way analysis of variance (ANOVA) revealed that all 11 organoleptic parameters were statistically significant at a significance level of p < 0.0001 in product differentiations. Additionally, principal component analysis (PCA) demonstrated a robust positive correlation between the retail price and the organoleptic parameter “barrel aroma”, as evidenced by the results presented in

Table 7. Correlations between retail prices and sensory attributes.

Parameter	Retail Price
Retail Price	1
Color	0.0179
Hue	−0.4391
Fruity Aroma	−0.3055
Barrel Aroma	0.3617
Flavor Intensity	−0.1005
Acidity	−0.1053
Bitter	−0.3727
Astringent	−0.2127
Aftertaste	−0.2222

. At the same time, the retail price exhibited a negative correlation with the organoleptic parameters “fruity aroma” and “bitter”.

Interestingly, the retail price exhibited a negative correlation with both the bitter and fruity aroma parameters. These findings are visually depicted in Figure 1.

4. Discussion

In the basic training stage, both teams exhibited identical performance concerning the appearance parameters, specifically for the Ishihara test and the red wine color classification test. However, significant differences emerged concerning the olfactory and taste parameters. Team A outperformed Team B in the olfactory domain, while Team A also performed better than Team B in the taste parameters, albeit with minor discrepancies.

During the subsequent training evaluation, Team A demonstrated markedly superior results compared to Team B. Notably, the disparity in olfactory performance between the two groups was much greater than their differences in taste parameters.

The findings from the basic training and evaluation tests highlight the advantage of candidate assessors participating in an accredited olive oil organoleptic evaluation panel. Members of such panels undergo a rigorous selection process distinct from that of non-participating testers [34]. This difference in performance can also be attributed to the ability of candidate assessors from accredited panels for olive oil evaluation to concentrate better during assessments, as opposed to new candidate assessors.

Interestingly, the similar performance of Teams A and B in terms of visual parameters can be explained by the fact that olive oil organoleptic evaluation does not incorporate any visual attributes [32]. In order to eliminate any potential psychological influence from sample appearance, an appropriate glass is used to conceal the color of the olive oil [46]. Conversely, the focus of olive oil organoleptic evaluation lies in olfactory parameters, which accounts for the substantial performance gap observed between Team A and Team B in this domain.

While candidate assessors from an accredited panel for olive oil evaluation generally excelled, some individuals from Team B outperformed their counterparts in Team A during the training assessments. Specifically, two Team B candidate assessors surpassed two tasters from Team A. Consequently, the selection process for sensory wine analysis (Panel C) included eight assessors from Team A and two from Team B. The unselected tasters formed Panel D, continuing the tests to evaluate the reliability of the selection method.

The evaluation of panel performance and individual taster performance was based on three key sensory attributes: a defect, an odor, and a taste attribute. These parameters were selected following the philosophy of the organoleptic evaluation of olive oil, where a defect and fruity attribute are used to monitor the team and taster performance [35].

This study’s results unequivocally demonstrate the appropriateness of the assessor selection process. Panel C outperformed Panel D by a significant margin. Although Panel D had several results falling outside the acceptance criteria, it performed slightly better for the taste attributes than the olfactory parameters. Notably, Panel C exhibited a remarkably low relative standard deviation compared to other studies involving trained and untrained panels [24], providing additional evidence of effective training, evaluation, and final selection.

In the context of Nemea wine samples, the sensory analysis identified five samples with aroma defects. These effects were estimated to be of low intensity, as their values were below 3.5 [32]. Two samples exhibited an oxidized defect, possibly resulting from chemical aging or storage conditions. Another sample showed a reductive defect, likely arising from random errors introduced by a bottle or batch. Additionally, two samples exhibited an animal defect, which was confirmed through GC-MS/MS analysis. Previous research has demonstrated that the presence of 4-ethylguajacol, 4-ethylphenol, and 4-vinylphenol contributes to the phenolic, animal, and stable characteristics of wine [47]. The detection of 4-ethylphenol, 4-ethylguajacol, and 4-vinylphenol in both samples suggests potential Brettanomyces growth. Specifically, Brettanomyces bruxellensis can produce 4-ethylguajacol from ferulic acid, as well as 4-ethylphenol and 4-vinylphenol from p-coumaric acid [48]. Furthermore, the confirmation of the organoleptic defect via the GC-MS/MS results underscores the effectiveness of the panel’s training in identifying specific defects.

In seven wine samples, the barrel aroma intensity surpassed the fruity aroma intensity. Interestingly, the average retail price of all the surveyed products was EUR 11.8, but these seven specific products commanded an average retail price of EUR 16.7, representing a substantial 42% increase.

Principal component analysis (PCA) revealed that barrel aroma exhibited the only positive, strong correlation with retail price. Simultaneously, fruity aroma and bitter demonstrated a negative correlation with retail price. The interpretation of this specific correlation centers around the aging process of wines in oak barrels. Evidently, this particular aging process leads to an increase in barrel aroma, a decrease in fruity aroma, and a reduction in bitterness, attributed to the polymerization of tannins. The specific pricing policy is explained by the fact that barrel usage results in increased production costs, which are then passed on to consumers through higher retail prices. Interestingly, our findings diverge slightly from other studies conducted in different countries, where higher astringency, fruity character, and oak influence were associated with more expensive products [5].

From the one-way analysis of variance (ANOVA) of the sensory descriptive data for the 25 wines, it was found that all 11 attributes differed significantly across the wines. More specifically, all 11 organoleptic parameters exhibited a p-value of less than or equal to 0.0001 (p ≤ 0.0001). This result underscores the correct choice of sensory attributes and the panel’s adequate training. Specifically, only two olfactory parameters—fruity aroma and barrel aroma—were chosen. This choice aligns with previous studies, which highlight that humans have limited ability to evaluate the intensities of various odors within complex aroma compositions [24,49]. For the same reason, in 1992, the International Olive Council (IOC) revised its method to enhance the consistency of panel assessments, limiting the number of olfactory parameters in fruitiness and aroma defects [50].

5. Conclusions

In our current study, a sensory analysis method for wine was developed that encompassed four essential components: (1) the training of candidate assessors, (2) the selection of assessors, (3) a method for sensory analysis, and (4) a methodology for monitoring tasters’ performance and the panel’s performance.

The method for the selection of assessors was demonstrated to be highly effective based on the extremely satisfactory results obtained during method validation. These results were characterized by a very low relative standard deviation while simultaneously meeting the criteria for precision number and deviation number.

The analysis method involves attributes that exhibit significant differences from one another, as demonstrated through the one-way analysis of variance (ANOVA) tests.

Additionally, the findings of this study revealed that candidate assessors who actively participate in an accredited olive oil sensory evaluation panel exhibit superior performance during the training process. Consequently, they have a significantly higher likelihood of being selected for the final sensory analysis panel compared to candidate assessors who did not engage in olive oil sensory evaluations.

Lastly, the present investigation established a robust positive correlation between the retail price of Nemea wine products and the sensory attribute of barrel aroma while also identifying a negative correlation between the sensory attributes of bitter and fruity aromas.