1 1.111I 01:11? {1:1 :4'p'j1 .11.;‘I1: ‘1 I. III}; 11%;;1; It 11131111111111 IH‘ In 11 11M; I :11 _-I ..1 .1” _‘ v _ F...“ 0.....- v—_.~ -.._“ -._ upmb': .. .— 3:1.‘- m. “M .— . ‘ _...‘a __ - -' _ = ; ' c *- _ ‘ 5-- h ' W ‘. m. 1— y,— —- - -'-—- as. _.._. M 4-“... '7‘. *.,~__"‘_ _ b .q«_‘ ”‘73... 111111111 III ' II'I 1. ‘1 III 1111 I111.I1111 11'1111111II : 1'1"”11‘ 11I1111111111111 I I1 "I' IIII'II‘I'II J I I101“ IIIQLI1 III 111‘: II IIIIIIII. 1 I. I I 1111111111111111111I1111IIII I1IU LII 11111 1111.11 I11“ 1II1I1‘1‘II1311‘I, t’“ I" : . 1.1 _> 11111I1II1lI'11I1r 1I1111 fi'la'11IItII1‘I” I11 1’11'1 1111 I171“! zI 1F11'II I111131-1‘5'1 1“ 1 1"111111111111111111111111111111111111111I1D'1'111 111I1 I1 1 II1 111111111 4.1I' 111111111411: 41-1“ 3::1"~1::§I z-‘ :I ‘ I 1- .11 11:'1 1 111111111 11111111111111.1I1‘11111111I 111111( I 1I1r1'1311111I21‘ 111114211111131155.11.1111‘1‘11 11 '1 11 11 1. 1'1 $111 I12! M1? {1111: I11! 1I11'1111.1 I‘IIII I11‘II11I11 11"1' 1111111 'i131.~11"11° 111' ' " ' I I.” ‘1' 11111111111111 1W1 I111I111111111II1I 111: 1,.III III: .21 II” IIII I I111I11 I ‘ , 11' . 5 11.:1I11 1111'11IIII11111 1111 1 I111 1111 11 111.“ ...... H131 :11 I :II I“ I :II III'III'III;w " .4 1I11II11111111111111111 111'11 ‘ . '11; 1 1111‘ IIIII11I1111351111111111111111’11111111111111111111 I £111111I1I1I‘”;11 11111 1"1'1 J111111 11111 111111 111 1111111 .111111 1I1IIIII1 1'111 11I1 1 ”11111111 ' 11’1111II'1I I'I1l III11 .IIIIII 111111611 111'II111 \ IIIII 1'11]? 1I1I11 ' I III11111' 11' IIIIUII “1| HIIHIIIII IIII1 M ~. I; I III III 1I111111111'II1I1IIII1:.I1III11111111 III‘U I.‘II'III*'=~~“I.. I-a II'IIIII ."1 1111', 1‘111I'11" {1‘ 1 1; ' II 1.1111" 4'“! IIIJIIirI 11I11 II? p111|1t1195II1 llllllllllIlllllllllllllllllllllllllllllllHIHHIUIIIOLHIUI 9/ 3 1293 104 9 “than— J— m. i LIBRARY Lw"‘e's:"y J watery-r 1'- “VJ—— This is to certify that the thesis entitled The Study of Aroma Characteristics of Raw Carrots With the Use of Factor Analysis presented by Mark Richard McLellan has been accepted towards fulfillment of the requirements for PhD degree in Food Science / %r professor Date April 24, 1981 0-7639 OVERDUE FINES: 25¢ per day per item ’ammfi ugmv MATERIALS: Place in book return to remove charge from circulation records The Study of Aroma Characteristics of Raw Carrots With the Use of Factor Analysis By Mark Richard McLellan A DISSERTATION Submitted to Michigan State University in partial fulfillment of the requirements for the degree of DOCTOR OF PHILOSOPHY Department of Food Science & Human Nutrition 1981 // i”. /, 97 ABSTRACT The Study of Aroma Characteristics of Raw Carrots With the Use of Factor Analysis By Mark R. McLellan Raw carrot aromatic volatiles were collected and concentrated using the porous polymer, Tenax-GC. Some of the collected volatiles were identified using mass spec— trometry and by comparing retention times with those of standards. Ratios of headspace volatiles were quite different from those previously reported in the literature, due to the mild collection conditions used in in this study. Sensory evaluations on twenty aroma-related characteristics were performed on the raw carrot aroma of ten largely different carrot selections. The sensory evaluation data were collected through the use of a micro- computer built into the sensory evaluation booth. Factor analysis was applied to the sensory evaluations to deter- mine true orthogonal descriptors which were derived from the original set of data. Five distinctly different characteristics of the raw carrot aroma were determined: (1) an earthy organic aroma, (2) a basic raw carrot aroma, (3) the intensity level of aromatics other than carrot, (A) the desirability level of pleasent aromatics (non-earthy) in carrots and (5) piney aroma. These aroma characteristics constitute an accounting of some 70% of the variation in raw carrot aroma. Through factor analysis, these new aroma characteristics were reconstructed in the form of a new data base to be evaluated with the headspace analysis data. A second application of factor analysis was utilized to point to possible peaks in the headspace chromatographic profile which were tested using multiple regression techniques to determine relationships to the five new sensory charcteristics. Only one of the regres- sion equations had an indication of significance and it related to the earthy organic aroma and three trace level peaks. When further tested for inclusion of the regres- sion variables in the equation, none proved significant. These findings support the contention that the aroma constituents of raw carrots play a minor role when compared to the taste parameters in the acceptance of raw carrots. THIS MANUSCRIPT IS DEDICATED TO THE THREE PEOPLE I HOLD CLOSEST TO MY HEART My wife - Deanne My son - Justin AND My Dad 11 ACKNOWLEDGMENTS I am sincerely grateful to Dr Jerry N. Cash for his suggestions and support throughout the course of my educa- tion at MSU. As my major professor, he was a source of both guidance and friendship. I would like to also thank the other members of my guidance committee, Drs. L.R. Baker, J.I. Gray, G.L. Hosfield, R.C. Nicholas, and M.A. Uebersax for their time and interest shown in this study. Special thanks is extended to the Carrot Breeding Program at MSU and Dr Larry R. Baker for providing the varous raw carrot selections used in the study and to Arun Mandagere,for his assistance in mass spectrometry. Appreciation is also extended to my fellow colleagues in their degree programs at MSU for their suggestions and support. Finally, I am most grateful to my wife for her limitless support and love which kept me going with a smile when I needed it most; and to my first child, Justin, who added an extra Joy to my life in these last long months. 111 TABLE OF CONTENTS LIST OF TABLES O O O O O O O O O O O 0 LIST OF FIGURES O O I O O O O O O O O 0 INTRODUCTION 0 O O O O O I O O O O O 0 LITERATURE REVIEW 0 O O O O O 0 O O O 0 Carrot Flavor Volatiles . . . . Carrot Taste . . . . . . . . Carrot Storage. . . . . . Sensory Descriptors of Carrot Flavor . Computer Interfacing to Sensory Measureme 50000 d U) MATERIALS AND IIETHODS . O O O O O I O I 0 Selection of Raw Carrots for Study Environmental Storage Chamber. . Porous Polymer Trap Preparation . Sampling and Volatiles Collection Trap Elution . . . . . . . Gas Chromatography . . . . . Mass Spectrometry . . . . . "30000... i Sensory Evaluation - Open Discussion P of le Panel . . . . . . . . . Sensory Evaluation - Discrimination / Intensity Testing and Training . . . . . . Sensory Evaluation - Raw Carrot Analysis . . . Statistical Analysis. . . . . . . . . . RESULTS AND DISCUSSION. . . . . . . . . . Experimental Design . . . . . . . . Controled Environment Chamber. . . . . Selection of Carrots. . . . . . . Porous Polymer Traps and Collection System Gas Chromatography . . . . . . . . iv Page vi ix 11 1H 16 17 17 19 19 2O 21 21 23 25 25 27 28 30 3O 33 33 3A 35 RESULTS AND DISCUSSION (cont.) Peak Identification . . . . . Sensory Evaluation . . . . . Modified Open Discussion Panel . Testing-Training Panels. . . . Qualitative Descriptive Analysis. . Factor Analysis . . . . . . . Computer Software for Sensory Analysis CONCLUSIONS . . . . . . . . . . . APPENDICES O O O O O O O O O O O A. B. C. D. E. F. G. H. I. J. K. L. M. N. LIST Volatile Constituents of Carrots . . Relative Humidity Control System . . Volatile Collection System . . . . Porous Polymer Trap Elution System. . Gas Chromatography . . . . . . . Aroma Standards . . . . . . . . TBS-80 Mod II Microcomputer System. . Personalized Ballots . . . . . . Sensory Evaluation Booth Construction. Data Collection Program . . . . . Mass Spectrum of Some Compounds in the Headspace of Raw Carrots . . . Calculations for the Coefficient of Concordance " W ". . . . . . Motivational Literature. . . . . . Factor Analysis Results . . . . . OF REFERENCES . . . . . . . . 75 81 83 89 93 106 110 115 115 119 120 121 122 126 128 130 131 132 1A1 1A9 150 153 158 Table 10. LIST OF TABLES Page Carrot lines and cultivars utilized in the analysis of raw carrot volatiles and for sensory evaluation of aroma attributes ....... 18 Gas chromatographic data for aroma volatiles eluted from a porous polymer trap for carrot selection MSU-1:413. 0.000000000000000000000000 38 Gas chromatographic data for aroma volatiles eluted from a porous polymer trap for carrot seleCtion MSU-1385 0.0.0.0.0000...0.0.0.000... ’42 Gas chromatographic data for aroma volatiles eluted from a porous polymer trap for carrot 86160t10n MSU-1383 coco.00000000000000.0000... “5 Gas chromatographic data for aroma volatiles eluted from a porous polymer trap for carrot 86180151011 MSU-5987 000000000000000000000000... ’48 Gas chromatographic data for aroma volatiles eluted from a porous polymer trap for carrot SGleCtion MSU-107 coooooooooooooooooooooooooo 53 Gas chromatographic data for aroma volatiles eluted from a porous polymer trap for carrot seleCtion SpartansweetOOOOOOOOOOOOOOOOOOOOOOOO 53 Gas chromatographic data for aroma volatiles eluted from a porous polymer trap for carrot seleCtion Spartan Fancy O. O... 0.0 00.00.. .0... O O 57 Gas chromatographic data for aroma volatiles eluted from a porous polymer trap for carrot selection GOldpaKOO...OOOOOOOOOOOOOOOOOOOOOOOOO 62 Gas chromatographic data for aroma volatiles eluted from a porous polymer trap for carrot selection MSU-6000............................ 62 vi Table 11. 12. 13. 1A. 15. 16. 17. 18. 19. A1. A2. A3. AA. A5. Gas chromatographic data for aroma volatiles eluted from a porous polymer trap for carrot selection Gosinoostovakja..................... List of trace level peaks detected in the porous polymer trappings of the raw carrot headspaceOOOOOCO0.0000.00.00000COOOOOOOOOOOOOO Mean squares and degrees of freedom as analyzed using one way analysis of variance... Compounds identified through the use of gas chromatography and mass spectrometry of the volatile constituents trapped on the porous polymer, Tenax-GC............................ Mean squares and degrees of freedom as analyzed using one way analysis of variance for the qualitative descriptive analysis data................................ Summary of data variation explained in the factoring of the sensory evaluation data..... The rotated factors for the factor analysis of the sensory evaluation data using the varimax criteria system of rotation.......... Equations used for the calculation of new faCtor VariableSOOOO000.00.00.00.00.0.0000... Multiple Regression Analysis of prediction equations for each factor variable involved in the second factor analysis................ Volatile constituents of carrots identified by BUttery 81: al., 1968’ 1978, 19790000000000. Volatile constituents of carrots identified by Buttery et al., 1968, 1978, 1979........... Volatile constituents of carrots identified by Murray and Whitfield, 1975................. Volatile constituents of carrots identified by Cronin and Stanton, 1975................... Volatile constituents of carrots identified by Linko et al., 197800.00000000000000000.0... vii Page 65 66 68 76 91 99 99 102 105 115 116 117 117 117 Table Page A6. Volatile constituents of carrots identified by Simon et al., 1980.00000000000000IOOOOOOOOO 118 E1. Average peak area data for all carrot selections included in the aroma study........ 125 F1. The ingredients of the standard aromas used in the training/testing of panelists.......... 126 F2. List of definitions made available to the panelists for both the testing/training panel and the carrot analysis panel.................. 127 N1. Correlation coefficients for the factor analysis of sensory evaluation data............ 152 N2. The factor matrix using alpha factor for sensory evaluation data........................ 15A N3. Explained variation for new sensory data and peak area used in the second faCtor' ana1y818000000000OOOOOOOOOOOOOOOOOOOOOO. 15“ NA. Varimax rotated factor matrix for new sensory data (previous factor variables) and peak dataOOOOOOOOOOOOOO0.00.00.00.00..00... 155 viii LIST OF FIGURES Figure Page 1. A continuous extraction apparatus first designed by Liken and Nickerson (196”) for the extraction of volatiles with a comparatively small amount of solvent......... 5 2. 3-sec-butyl-2-methoxypyrazine................. 10 3. 3-Methyl-6-methoxy-8-hydroxy-3,u- dihydroisocoumarin............................ 13 A. Gas chromatogram of carrot selection MSU-1A13 in the Raw Carrot Aroma Study........ 37 5. Gas chromatogram of carrot selection MSU-1385 in the raw carrot aroma study................. No 6. Gas chromatogram of carrot selection MSU-1383 in the raw carrot aroma study................. A3 7. Gas chromatogram of carrot selection MSU-5987 in the raw carrot aroma study................. A6 8. Gas chromatogram of carrot selection MSU-107 in the raw caPPOt aroma Study. 0 O O O O O O O O O O 0 O O O O “9 9. Gas chromatogram of carrot selection Spartan Sweet in the raw carrot aroma study........... 51 10. Gas chromatogram of carrot selection Spartan Fancy in the raw carrot aroma study........... 55 11. Gas chromatogram of carrot selection Gold Pack in the raw carrot aroma study............ 58 12. Gas chromatogram of carrot selection MSU—6000 in the raw carrot aroma study................. 60 13. Gas chromatogram of carrot selection Gosinoo— strovakaJa-13 the raw carrot aroma study...... 63 1A. Graphic presentation of the means for peak #7 for all the carrot selections included in the studyOOOOOOOOOOOOOOOOOOOOOOOOOOOO0.0.0.0000... 71 ix Figure Page 15. Graphic presentation of the means for peak #13 for all the carrot selections included in the StudyOOOOOO0.0.0.0....00......OOOOOOOOOOOOOO... 71 16. Graphic presentation of the means for peak #21 for all the carrot selections included in the stUdyOOOOOOOOOOOO0.0.00.0.0.0000...0.0.0.000... 73 17. Graphic presentation of the means for peak #25 for all the carrot selections included in the studyOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOCOOOOOOOO 73 18. Schematic of the Stimulas - Responce Circuit typically found in man.’OOOOOOOOOCOOOOOOOOOOOOO 82 19. Histogram for sensory aroma characteristics generated from the Open discussion panel....... 8A 20. Summary of the Qualitative Descriptive Analy81SOOOOO0.0.0.00..0000.0COOOOOOOOOOOOOOOO. 90 21. Qualitative Descriptive Analysis model for sensory attributes of carrots based upon mean values for each descriptor..................... 92 22. Summary of the interpretation of the factor analysis applied to the sensory evaluation dataOCOOOOOOOOIOOOOOOOOOOOOOOOOOOOOOOO0.0.0.... 101 Bl. Schematic of the control system for maintanence of the relative humidity in the environmental storage chamber. ................ 119 C1. Schematic of the volatile collection system using the porous polymer traps and nitrogen sweep teChnlqueOO0.0.0...OOOOOOOOOOOOOOOOOOOOOO 120 D1. Schematic of the trap elution mechanism........ 121 H1. The personalized ballot produced for each panelist in the testing/training phase of theStUdyoooooooooooooooooooooooooooooooooooooo 130 11. Schematic of the sensory evaluation booth constructed around the microcomputer system for use in the sample evaluation phase of the study...................................... 131 K1. Mass spectrum of peak #1 with the molecular ion denoted as "M"0.00.0000...OOOOOOOOOOOOOOO 1112 X Figure K2. K3. K14. K5. K6. K7. Ml. Mass spectrum of ion denoted as " Mass spectrum of ion denoted as " Mass spectrum of ion denoted as " Mass spectrum of ion denoted as " Mass spectrum of ion denoted as " Mass spectrum of ion denoted as " Page peak #2 with the molecular M"OOOOOIOOOOOOOOOOOOOOOOOOO0.. 1H3 peak #u with the molecular M"O0.00.00.00.00...0.0.0.0.... 114“ peak #5 with the molecular M"O0.0.000...OOOOOOOOOOOOOOOOO 1&5 peak #9 with the molecular M."OOOOOOOOOOOOOOOOOOOOO0...... 1H6 peak #11 with the molecular p1".0...OOOOOOOOOOOOOOOOOOOIOOO 1147 peak #13 with the molecular Mv'.C.OOOOOOOOOOOOOOOOOOOOO0.0. 1&8 A newsletter distributed to the panelists as a motivation for achievement................. 150 xi INTRODUCTION Our lack of knowledge concerning the mechanisms by which we perceive tastes and odors contrasts starkly with our understanding of the processes by which we preceive sounds and visual images. In general, we can store, retrieve, amplify, transmit, duplicate, and describe objectively the sights we see and the sounds we hear. Unfortunatly, few if any of these operations can be duplicated for a single taste or odor. For man, flavor and nutrition are simply two sides of the same coin. What smells and tastes good is eaten with sensual delight, many times irregardless of nutritional benefit. Man will reject the most nutritious and wholesome food, unless he got some degree of olfactory and sapid enjoyment. In his attempt to understand the complexities of flavor, man has turned to the analytical chemist for insight into the chemical makeup of our food. We have asked the sensory analyst to evaluate our Judgements on flavor, interpeting our responses as measurements of gustation and olfaction. Physiologists study the construction and "operation" of our flavor senses and the 2 behavioral-psychologists attempt to interpret the thought“ processes behind it all. Man has thrown the full weight of his scientific advancement against the barriers of ignorance preventing our understanding of these special senses. And so far, comparatively little progress can be shown for it all but studies continue investigating the relationships between the chemistry of food and its sensual characteristics. Carrots, (Daucus carota EL) are a relatively important crop in the United States and are used extensively for fresh market and processing into canned, frozen, and dehydrated products. The Division of Economic and Statistical Analysis of USDA reports that including fresh, frozen, and canned carrots the population of the United States consumes some 9.6 grams of carrots/ capita/ day (Powel, 1981). Improving the culinary quality of raw carrots appeals to both the processor and consummer alike. Various phases of this research have been attempted with differing degrees of success. In most previous studies that involved sensory analysis of carrots, panelists were asked to taste the carrots and rate various attributes. These attributes varied; however, most measured the sapid attributes of the carrot tissue. Any aroma characteristics were for the most part measured using very general terms such as "flavor intensity" (Schreerens and Hosfield, 1976), "overall flavor and harsh flavor" (Simon et al., 1980). And even in these cases, the ability to distinguish between taste and smell were not strived for. In this study, an attempt was made to evaluate the aroma attributes of raw carrots. No taste parameters were included. Sensory evaluations of the aromas were compared to objective measurements of aroma volatiles with the use of factor analysis. LITERATURE REVIEW A. Carrot Flavor Volatiles. Extensive studies have been done on the composi- tion of carrot seed oil (Seifert et al., 1968), but until 1968, virtually no studies were done on the composition of volatiles present in the carrot root. Buttery et a1. 1968, published the results of their in-depth analysis of carrot root volatiles utilizing a steam volatile oil col- lected at atmospheric pressure with a continuous extrac- tion apparatus as shown in Figure l (Liken and Nickerson, 196h). The apparatus allows simultaneous condensation of the steam distillate and an immiscible extracting solvent. The distillates return to their respective distillation flasks via arms at different levels. In Figure l, the water phase returns through arm B and the low boiling alkane through arm A. The steam can be introduced into the system by heating the carrot slurry or by introducing it from an outside source . In the latter case a water draw off valve and drain is included in the design. The low boiling alkane must be heated to maintain a relatively high vapor pressure for the extraction process. For the Figure 1. A continuous extraction apparatus first designed by Liken and Nickerson (196M) for the extraction of volatiles with a comparatively small amount of solvent. Some discussed modifications are presented. (Lester,1981) ”CO 34cc H— .Pz w>40m H 02;0 Area Reject to 5 Time 6.00 > Attentuation to 1 Time 10.00 => Rate to 0.01 Time 1H.00 > Rate to 30.00 Quantitative results for the purpose of compari- sons between various carrot sample were based on reported peak area. Comparisons were considered valid as carrot samples were treated identically. Five standards were purchased (K&K Fine and Rare Chemicals, Plainview NJ.) for use in the identification of some of the chromatographed volatiles : alpha-D—Pinene 95% (Lot-38117A), Myrcene Techn. (Lot—31123-A), DL- Limonene (Lot-31123-A), Isobornyl Acetate (Lot-32075-A), and Beta-Carophylene Techn. (Lot-3N2h9-A). G. Mass Spectrometry. Fifteen traps were collected of a carrot selection mixture that would provide most of the compounds for high resolution mass spectrometer analysis. The eluted com- pounds were concentrated approximatly 20 fold through a an endothermic nitrogen evaporation process. The resultant 2A concentrate was used in the HP 5989 GC-mass spectrometer system. The column used in the GC/MS system was a six foot glass column of Amm id and packed with 5% Carbowax 20M. The following conditions were maintained by the gas chromatograph for the packed column: Detector : Total ion Carrier Gas : Helium = A0 ml/min. Temperature - Time Microprocessor Functions: Temperature 1 = 300 Time 1 = A.0 minutes Rate = A.00/minute Temperature 2 = 1700 Time 2 = 10 minutes Injection temperature = 150C 0.75 cm/min. Chart speed 10 Attenuation Area reject = (-) Time A.5 => Attenuation to 6 Time 5.0 => Area reject to 1 Time 12.0 => Rate to 6C/min The mass spectrometer was set up for a run time of A0 minutes. Start - Stop masses were no and A00 m/e respec- tively. Analog to digital conversion was on the order of 3 measurements per minute. The scan start was set to a 30 second delay with a threshold of 5.0, the most sensitive setting possible in relation to signal to noise ratios. The ion source temperature was set for 2000. The electron 25 multiplier was adjusted for 2200 electron volts and the ion source emmited 70 electron volts for standard ioniza- tion. H. Sensory Evaluation - Open Discussion Profile Panel. Prior to panel testing and sample evaluation an open discussion panel was convened for the purpose of developing proper descriptors for raw carrot aroma. Two sessions were run where each of seven judges was given two sets of five samples labeled A through E and F through J. The profile panel consisted of three stages : (1) initial discussion, (2) private evaluation and (3) open discus- sion. Each set of samples was presented to the panelists with a general discussion of the raw carrot puree. Panelists were asked to privately characterize the aroma by writing down appropriate descriptors. Open discussion of each sample was subsequently encouraged to exchange ideas and impressions. Judges were allowed to add to their list of descriptors during the discussion. Finally, each judge was asked to rank the five samples in order of preference based upon aroma. This process was repeated for the second set of samples and again for the two sets of samples in the second panel. I. Sensory evaluation - Discrimination / Intensity Testing and Training. All panelists were introduCed to nine standards (Appendix F, Table F1) representing each of the classes of 26 smells identified by the flavor profile panel. No attempt was made to exactly duplicate any one odorant identified in carrots but rather the classes of odorants were repre- sented. All samples were placed in plastic containers covered on the outside so that nothing inside could be seen and the sample was also covered on the inside with a A cm thick pad of tight fitting Dacron Fiber Fill II. All sample cups remained covered with lids until uncovered by the panelists. The panelists had a list of aroma definitions before them at all times during the sniffing (Appendix F, Table F2.). After the introduction session, three testing sessions were run where each panelist was required to properly identify the nine standards. A data base was ~ maintained on a TBS-80 Model II microcomputer (Tandy Corp., TRS-80 Model II , Appendix G, hereafter simply noted as "the microcomputer") for each panelist containing their scores on the testing/training phase. Each panelist was given a personalized ballot (Appendix H, Figure H1) containing not only the questions but also that panelist's previous results. The test sessions were started two days after the introduction session and ran for the next two days with the third panel following two days after the end of the first two test days. Panelists were allowed to get one answer wrong on the first test panel; however, thereafter all panel tests had to be completely correct. Panelists correctly completing all requirements 27 were used in the sensory analysis of the raw carrot samples. J. Sensory Evaluation - Raw Carrot Sample Analysis. Carrots were prepared as explained in preparation of raw carrots for volatile analysis. A covered plastic cup was prepared for each sample, for each panelist. Approximately 50 ml of carrot puree were put in each cup and covered with a tight fitting cap until opened by the panelist. Sensory analysis of carrot samples was performed using the microcomputer as the medium of interaction and data acquisition with the panelists. The panel booth was constructed around the microcomputer so that the integra- tion of the two was uniform and non-distracting (Appendix 0, Figure 01.). All questions asked of the panelists were presented on the video screen of the microcomputer through the use of a specially designed program writtern in BASIC (An acronym for Beginners All-purpose Symbolic Instruction Code) for the microcomputer (See Appendix I, Figure Il). Panelists initiated the session by typing in their last name on the microcomputer console keyboard. Once initiation of the panelist was completed for that session, a specially constructed mask was placed over the keyboard which only enabled access to the ">", "<", "?" and "space bar" keys. These keys represented all that was required to complete any acceptable response from the 28 panelists. Although each session consisted of 3-u samples, the order of sampling was randomized by the computer and this order of sniffing was relayed to the panelists via messages on the video screen. Responses were automatically measured, interpreted, coded and stored for later use in statistical analysis. K. Statistical Analysis. In the open discussion profile panel the frequency of all responses were measured. An arbitrary inclusion point was chosen to be a frequency of 50% of the partici- pating judges, hence an characteristic having a recorded frequency over 6 was considered for use in the study. The ranking that was done for the two sets of samples in the profile panel was analyzed using two methods of analysis. To determine if there was a signifi- cant agreement between rankings assigned to the samples by the judges; an analysis of variance technique was applied using a Special statistic W, the coefficient of concord- ence (Kendall, 19A8). To determine if a sample was significantly different from others in the group of similar samples on the basis of ranking, Kramer's Rank Sum Method was used (Kramer, 1960). Variation between carrot samples for peak areas was analyzed using an analysis of variance by peak for all carrot samples. Variation between carrot samples for sensory eval- 29 uation parameters was analyzed using an analysis of vari- ance by sensory parameter for all carrot samples. To evaluate the existence of some underlying pat- tern of relationships in the sensory evaluation data a factor analysis was implemented. The factor analysis evaluated the pattern of relationships to identify and interpolate a set of source variables accounting for observed interrelations in the data. A calculation of new source variables was made with the data to enable the application of a second factor analysis including peak area data for the evaluation of underlying relationships. These relationships were tested for levels of significance using multiple regression analysis. RESULTS AND DISCUSSION Experimental Design. The carrot selections were chosen on the basis of a carrot breeders' written comments in field trial records. The selection process strived to produce as widely varied a sample as possible. The more the variance on this level the greater the power of study in determination of differ— ences later. Carrot seed for this study were planted on muck soil in a completely randomized block design with dupli- cate rows per selection. Sample preparation for gas chromatography was designed to eliminate intra-varietal differences of aroma volatiles. Fifteen to twenty roots of each selection were trimmed, sliced and a 1000g sample was blended for use in trappings and aroma panel studies. Collection and trap- ping for gas chromatography samples included three sub- samples of 100g each, connected to a porous polymer traps. Two open discussion panels were run in order to generate proper descriptors for use in the characteriza- tion of the raw carrot aroma. Panelists were directed to select and list privately chosen descriptors. Following 31 the private evaluation, an Open discussion was promoted where panelists were encouraged to discuss each carrot selection and to develop a uniform description of the aromas. Panelists were encouraged to include descriptors from the discussion in their listings where appropriate. This method differs from the classical approach to discus- sion profile panels (Amerine et al., 1965) where only the descriptors from the Open discussion section are included in the analysis. In the approach used in this study, the results include not only open discussion descriptors but also the individual descriptors. As a final step the descriptors totaling to 50% of the total number of panel- ists were included in the final list of descriptors. This approach ensures, the collection of descriptors not discussed, yet written frequently, as well as those de- scriptors that were discussed openly by the panel. The testing-training phase of the study was designed to accomplish the task of preparing panelists for separating and identifying classes of compounds in raw carrot aroma. It also served to eliminate panelists, based on their performance in the areas of sensitivity, separation of odor character, and consistancy. The Qualitative Descriptive Analysis was designed as a completely randomized block with duplicates. The factor analysis was included in the design to accomplish an evaluation of underlying trends and reduc- tion of sensory data into orthogonal parts. The results 32 of this first analysis were used for calculation of new factor variables for inclusion in a second factor analysis; which included both new factor variables and peak area data. The approach chosen in the design of this study is somewhat different from previous studies. It is realized from previous contributions to the literature that the levels of sweetness (soluble solids, sugars), bitterness (isocoumarin), and dry matter are of critical importance in the acceptance or rejection of carrots (Scheerens and Hosfield, 1976.; Carolus and E113, 1957.; Carlton and Peterson, 1963.). It was not the intent of this study to reaffirm the well supported findings concerning these attributes but rather break away from the repeated study of taste-flavor, the two not necessarily separable in this case, and initiate a study of the aroma of raw carrots. In this study, the word aroma is taken to mean smells detected by the nose without interference of taste stimuli. The raw carrot aroma is the smell of carrots in their raw state with no off odor due to heating. When eating raw carrots, the temperature at which the raw carrot aroma is smelled is no higher than body tempera- ture; hence the temperature for collection of headspace volatiles was set no higher than 300. 33 Controlled Environment Chamber The controlled environment chamber was set up to accommodate crates of carrots. Air flow was maintained at all times with a built-in circulating fan and moisture was introduced using the system described in the methods sec- tion. Relative Humidity was maintained between 95% and 99% . The electrical float control on the atomizer unit was found to develop corrosion problems after a year in operation. Repetitive cleaning of contacts remedied the situation; however, a replacement part where contacts were fully weatherized would be highly recommended. Although the control function in this system was simply a clock cycle gauged to the static environmental conditions of the chamber; an alternative system could be suggested where the control system included a Dunmore type hygrometer cell. In this system resistance of a sensing element varies with percent relative humidity (Ross, I.J., 1975). Selection of Carrots Based on Breeder's Comments. Previous year's field trial records for all carrot selections in the Michigan State University Carrot Breed- ing Program were evaluated for possible inclusion in the aroma quality study. The decision of whether or not to include a selection was made on a basis of the breeder's comment as to the eating quality of the carrot. It was realized that many of the descriptions and opinions were probably biased, however, they provided a base to start 3A from. Some of the descriptions recorded in the field trial records were : bitter, sweet, sweet pleasant smell, oily, carroty, strong, bland, perfumy, piney, and pleasant. Along with the selections chosen from the field trial books, several commercial varieties were also eval- uated. Porous Polymer Traps and Collection System. The collection system used in this study was a modified version of that described by Simon et al., (1980a). The packing weight was approximately the same as that used by Simon although the reported amount was some ten times greater than what was used in this study (Simon, 1981). The study of headspace volatiles offers several advantages to the aroma chemist: (a) A relatively small sample of food is required. (b) Very little sample preparation is required, therefore artifacts are kept to a minimum. (c) Compounds in the headspace are representative of what one actually smells. A limited concentration process will bring the headspace volatiles into the range of many analytical techniques (Teranishi et al., 1971). Of concern in this study was the possible development of the familiar off odor of cooked carrots, since the purpose of the study was to deal with the raw carrot aroma. In the determination of 35 collection conditions for this study, it was felt that even a holding temperature of 600 was too harsh because of the distinct off odor developed. It was concluded that after an hour of holding at 300, the collection tempera- ture of this study, no noticable off odor (cooked aroma) was detected. Gas Chromatorgraphy The elution time of volatiles off the porous poly- mer trap was quite fast, due to the use of a plastic manifold adapted to a syringe, which enabled moderate pressure to be applied to the solvent placed on the top of the trap. The chromatogram of the compounds eluted from the porous polymer trap was somewhat different from previous results (Simon et al., 1980a) because the collection con- ditions promoted a shift in the total peak area to the more volatile early eluting compounds. These included peaks 1 through 13 as indicated in a table of average peak areas (Table E1, Appendix E). In most previous works relatively moderate heating conditions were maintained which drove off the more stable higher boiling compounds. However, in this study particular care was taken to avoid any possible artifact formation or odor character degrada- tion. In Figure A a representative chromatogram of MSU- 1A13 is shown. The majority of the peak area is taken up by peaks 1 through 12 with peaks 1,A, and 5 being very 36 Figure A. -— Gas chromatogram of carrot selection MSU-1A13 in the raw carrot aroma study. 37 cu «N cw Nr 2 38 Table 2. -- Gas chromatographic data for aroma volatiles eluted from a porous polymer trap for carrot selection MSU-1A13. Peak No. Retengionrlngex Peak Area PercingaTogal 1 0.83 60A 17.517 A 0.92 765 22.187 5 1.00 1139 33.03A 8 1.13 111 3.219 9 1.16 355 10.296 11 1.2A 170 A.930 12 1.29 65 1.885 13 1.31 196 5.68A 22 1.83 10 0.290 23 1-97 33 0.957 '-(57 """"""""""""""""""""""""""""""""""""""" (b) Based upon Sabinene Excluding solvent peak 39 prominent. Table 2 is a breakdown of the gas chromato- graphy data. In this table, a retention index is used for comparison of all peak retention times to a standard (Heatherbell et al., 1971). Sabinene, which is peak 5, was chosen as a standard because it appears in each of the traps evaluated and falls towards the middle of the chromatogram. MSU-1385 is shown in Figure 5. Peak areas are quite high with over 59% of the total volatiles consisting of peaks 1,A, and 5 (Table 3). The carrot line MSU-1383 drops in amount of total volatiles overall but maintains the presence of some of the higher boiling compounds (Figure 6.). This would tend to indicate a higher ratio of high boiling compounds to low boiling compounds, otherwise one would expect a lower peak area for both high and low boiling compounds (Table A). MSU-5987 is shown in Figure 7. A large amount of total volatiles are present with five peaks registering with longer retention times than beyond peak 13 although peaks 1,9,and 13 constitute the majority of the volatiles (Table 5.). In comparison, MSU-107, is quite low in total volatiles (Figure 8) with only five peaks registering on the integrator (Table 6). Peaks 1,A, and 5 constituted the majority of the total volatiles. Figure 9 shows the commercial variety, Spartan Sweet, a fresh market product. Note that total volatiles A0 Figure 5. -- Gas chromatogram of carrot selection MSU-1385 in the raw carrot aroma study. 41 «.N «N 9N «up up or. we; N-O A2 Table 3. -- Gas chromatographic data for aroma volatiles eluted from a porous polymer trap for carrot selection MSU-1385. Peak No. Retentionrlngex Peak Area PercingaToBal 1 0.82 1292 16.3A6 2 0.85 70 0.885 3 0.87 99 1.252 A 0.93 1185 1A.992 5 1.00 23A1 29.617 8 1.13 211 2.660 9 1.16 908 11.A87 11 1.2A 331 A.187 12 1.28 190 2.A03 13 1.31 105A 13.33 1A 1.35 119 1.505 21 1.79 20 0.253 22 1.83 10 0.126 2A 1.97 7A 0.936 “(5) """""""""""""""""""""""""""""""""" (b) Based upon Sabinene Excluding solvent peak A3 Figure 6. -- Gas chromatogram of carrot selection MSU-1383 in the raw carrot aroma study. 44 #N A5 Table A. -— Gas chromatographic data for aroma volatiles eluted from a porous polymer trap for carrot selection MSU—1383. Peak No. Retentionrlngex Peak Area PerciggaTogal 1 ‘0.81 A95 33.559 A 0.92 337 22.8A8 5 1.00 AA8 30.373 9 1.17 97 6.576 12 1.2A 1A 0.9A9 22 1.78 11 0.7A6 2A 1.92 61 A.136 25 2.01 12 0.81A -'(§) -------------------------------------------- (b) Based upon Sabinene Excluding solvent peak A6 Figure 7. —- Gas chromatogram of carrot selection MSU-5987 in the raw carrot aroma study. ”13 47 fl A8 Table 5. -- Gas chromatographic data for aroma volatiles eluted from a porous polymer trap for carrot selection MSU-5987. Peak No. Retentioanngex Peak Area PerciggaTogal 1 0.80 2AAA 18.871 2 0.85 165 1.27A 3 0.86 120 0.927 A 0.92 AA6 3.AAA 5 1.00 586 A.525 7 1 08 A8 0.371 8 1.12 668 5.158 9 1.16 2027 15.651 11 1.2A 152 1.17A 13 1.31 6088 A7.008 20 1.7A 25 0.193 21 1.77 53 0.A09 22 1.83 78 0.602 2A , 1.96 30 0.232 25 2.05 21 0.162 ”-(57 """"""""""""""""""""""""""""""""""""""" (b) Based upon Sabinene Excluding solvent peak A9 Figure 8. -- Gas chromatogram of carrot selection MSU-107 in the raw carrot aroma study. 50 24 51 Figure 9. -- Gas chromatogram of carrot selection Spartansweet in the raw carrot aroma study. 53 Table 6. -- Gas chromatographic data for aroma volatiles eluted from a porous polymer trap for carrot selection MSU-107. Peak No. Retention I x Peak Area Percent T l tr/tr 733 Area ?B§ 1 0.81 126 21.687 A 0.92 169 29.088 5 1.00 ‘ 220 37.866 9 1.16 57 9.811 2A 1.93 9 1.5A9 ’“TET """"""""""""""""""""""""""""" (b) Based upon Sabinene Excluding solvent peak Table 7. -- Gas chromatographic data for aroma volatiles eluted from a porous polymer trap for carrot selection Spartansweet. Peak No. Retention I x Peak Area Percent T 1 tr/tr $g? Area ?83 1 0.81 110A A3.706 A 0.92 A87 19.280 5 1.00 666 26.366 8 1.13 37 1.A65 9 1.17 20A 8.076 2A 1.93 38 1.108 "'75) """""""""""""""""""""""""""""" (b) Based upon Sabinene Excluding solvent peak 5A appear low as compared to some of the previous lines shown (Table 7); however, as with virtually all of the previous selections, the majority of the total volatiles, over 80%, are found in peaks l,A, and 5. Spartan Fancy is also a commercial variety used for fresh market produce. It is interesting to note that Spartan Fancy, although a named variety, has very low total volatiles in comparison to previously mentioned breeding lines (Figure 10). The digital integrator interpreted only three peaks for integration and they were peaks l,A,and 5 (Table 8). The commercially grown U.S. variety, Goldpak, was included in the study. A representative chromatogram shown in Figure 11 indicates a majority of the volatiles were included in the three peaks l,A, and 5 (Table 9). In 1980, a study concluded that some breeding lines and cultivars had significantly higher levels of reducing sugars and non reducing sugars (Lester, 1980). Two breed- ing line from that study were included in this study of raw carrot aroma. They were MSU-6000, reportedly a high sugar line, and Gosinoostrovakaja-13, reportedly a low sugar line. MSU-6000 is shown in Figure 12 with support- ing data in Table 10. MSU-6000 was quite interesting because total volatiles werelow, but numerous high boil- ing compounds showed up in the chromatogram. Gosinoostro- vakaja-13, the low sugar line, appeared to be similar to MSU 6000 but with less high boiling compounds (Figure 13, Table 11). 55 Figure 10. —- Gas chromatogram of carrot selection Spartan Fancy in the raw carrot aroma study. 57 Table 8. -- Gas chromatographic data for aroma volatiles eluted from a porous polymer trap for carrot selection Spartan Fancy. Peak No. Retention I d x Peak Area Percent T l tr/tr 7a? Area $53 1 0.81 AA9 A3.677 A 0.92 323 31.A20 5 1.00 256 2A.903 "(5) """""""""""""""""""""""""""""" (b) Based upon Sabinene Excluding solvent peak 58 Figure 11. -- Gas chromatogram of carrot selection GoldPak in the raw carrot aroma study. 59 1O 24 60 Figure 12. -- Gas chromatogram of carrot selection MSU-6000 in the raw carrot aroma study. 61 22 19 17 62 Table 9. -- Gas chromatographic data for aroma volatiles eluted from a porous polymer trap for carrot selection GoldPak. Peak No. Retention I x Peak Area Percent T 1 tr/tr 53? Area 885 l 0.82 511 3A.786 u 0.92 387 26.3A5 5 1.00 A55 30.97A 10 1.17 9A 6.399 2A 1.91 22 1.A98 "I§T """"""""""""""""""""""""""""" (b) Based upon Sabinene Excluding solvent peak Table 10. -- Gas chromatographic data for aroma volatiles eluted from a porous polymer trap for carrot selection MSU-6000. Peak No. Retention I x Peak Area Percent T 1 tr/tr 5g? Area 58% 1 0.82 110 30.812 5 1.00 132 36.975 8 1.17 AA 12.325 17 1.63 9 2.521 19 1.69 52 1A.566 23 1.88 10 2.801 ““757 """"""""""""""""""""""""""""" (b) Based upon Sabinene Excluding solvent peak 63 Figure 13. -- Gas chromatogram of carrot selection Gosinoostrovakaja-l3 in the raw carrot aroma study. 24 65 Table 11. -- Gas chromatographic data for aroma volatiles eluted from a porous polymer trap for carrot selection Gosinoostovakja. Peak No. Retention I x Peak Area Percent T l tr/tr 593 Area 583 l 0.81 A87 39.085 A 0.92 365 29.29A 5 1.00 371 29.775 2A 1.97 23 1.8A6 "(57 """""""""""""""""""""""""""""" (b) Based upon Sabinene Excluding solvent peak 66 Withycombe et a1. (1978) were evaluating trace volatile constituents of hydrolyzed vegetable protein when they found that of a number of polymers tested, Tenax—CC produced the most organoleptically characteristic isolate. The Tenax-CC was compared to Chromosorb 105 and Porapak Q. Boyko et a1. (1978) reported that retention times of trapped compounds were shorter on the the Tenax-00 and found that during a twenty minute water removal step, losses of low boiling compounds would occur. Simon et al. (1980a) used Tenax-00 for the collection of carrot volatiles and reported an increasing variability in peak areas with the more volatile, low boiling compounds. Variability was found to be quite high in the present study, although measures were taken to minimize it. Water elimination was kept to a maximum of two minutes by nitrogen flush and all traps were capped with parafilm then held at -230 until elution. The variability problem was accentuated by the number of trace level components found in the samples (Table 12). Table 12. -- List of trace leveIa) peaks detected in the porous polymer trappings of the raw carrot headspace.(See Peak Identification p.75) Peak Number - Identity Peak Number - Identity Peak 6 - unknown Peak 7 - unknown Peak 1A - unknown Peak 15 - unknown Peak 16 - unknown Peak 17 - unknown Peak 18 - unknown Peak 19 - unknown Peak 20 - unknown Peak 21 - isobornyl acetate Peak 22 - beta-carophylene Peak 23 - unknown Peak 25 - unknown Peak 26 - unknown (a) less than an area of 15 on the digital integrator. 67 The term trace level volatiles was used to describe com- pounds not averaging over three times the integrator mini— mum peak height which was set at a peak area level of 5. Therefore, peaks averaging less than a peak area of 15 were considered trace level. In previous works on the volatile constituents of carrots, many compounds were found in large quantity but the conditions by which these compounds were collected could not be assumed to represent the raw carrot aroma. In this study, some compounds of the high concentration which previous researchers had driven out of the carrot puree by moderate to harsh holding conditions showed up as only trace quantities. A one-way analysis of variance was applied to the peak data to distinguish peaks which varied at a statistically significant level over the various carrot selections. Mean square values and degrees of freedom for each are shown in Table 13. As noted, four peaks were significantly different: peaks 7, 13, 21, and 25. Simon et a1. (1980) indicated that only three of the volatiles measured in their study were significantly different between carrot selections. They were alpha- phellanderene, limonene, and terpinolene. 0f the four peaks which varied significantly between carrot selections, peak 7 is unknown, peak 13 is gamma-terpinene, peak 21 is isobornyl acetate, and peak 25 is unknown. Testing of differences between means for the compounds was performed using Tukey's "Honestly 68 Table 13. -- Mean Squares and degrees of freedom as analyzed using One Way Analysis of Variance. Analysis of Variance for Peak Area Data Variable Between Groups Within Groups EFT-"$71?" 537'"??? P01 - alpha-Pinene 9 6376A1.7926 20 358753.6667 P02 — Camphene 9 96A.2370 20 1082.8667 P03 - unknown 9 699.5889 20 820.0333 POA - beta-Pinene 9 10591A.5037 20 109989.7333 P05 - Sabinene 9 38878A.A296 20 501897.8667 P06 - unknown 9 1.6333 20 1.6333 P07 - unknown 9 388.8000 20 388.8000* P08 - unknown 9 5A550.2259 20 26127.1667 P09 - Myrecene 9 5A3156.0037 20 269518.9333 P10 - unknown 9 3511.AA07 20 3210.7000 P11 - Limonene 9 9522.3000 20 A059.3333 P12 - unknown 9 26829.8667 20 29956.3333 P13 - gamma-Terpinene 9 55157A2.5519 20 1382385.5000* PlA - unknown 9 A66.8000 20 A7A.7333 P15 - unknown 9 13.3333 20 13.3333 P16 - unknown 9 108.1630 20 120.8667 P17 - unknown 9 28.6852 20 1A.5667 P18 - unknown 9 32.0333 20 32.0333 (continued) 69 Table 13. -- Mean Squares and degrees of freedom as analyzed using One Way Analysis of Variance. (Cont.) Analysis of Variance Peak Area Data Cont'd. =888:833:3=================338::2:833:=============8=====3=== Variable Between Groups Within Groups ”1337""?an 637"???" P19 - unknown 9 1303.6AAA 20 6A8.7333 P20 - unknown 9 15A.6519 20 197.2000 P21 - Isobornyl Acetate 9 602.0185 20 192.6667* P22 — beta-Carophylene 9 2A5.3370 20 192.8667 P23 - unknown 9 A9.070A 20 2A.1667 P2A - unknown 9 1361.2185 20 1258.7667 P25 - unknown 9 57.1889 20 18.8333* P26 - unknown 9 13.3333 20 13.3333 95% Significance Level 70 Significant Difference" test which utilizes a t-like statistic based on the distribution of the Studentized Range (Tukey, 1953). The test statistic used was: (‘7 - 7') ms / r 1. 1'. E with the critical value of :q equal to 5.008 for a,t,n-t, a = 0.05, t = 10, and t-n = 20. In Figure 1A, the means for the ten carrot selec— tions are shown for peak #7. The breeding line MSU-5987 is significantly different from MSU-1385, MSU-1383, MSU- 107, Spartansweet, Spartan Fancy, GoldPak, MSU-6000, and Gosinoostrovaka-13. MSU-5987 is not significantly different from MSU-1A13 for peak #7. Figure 15. shows the differences among carrot selection for peak #13, gamma—terpinene. MSU-5987 is significantly higher in gamma-terpinene than any other selection studied. Isobornyl acetate is peak # 21 and comparisons between carrot selections are shown in Figure 16. MSU- 5987 is higher in isobornyl acetate than any other selection studied. Peak #25 is compared between all of the carrot selections in Figure 17. MSU—5987 is shown to be significantly higher in this unknown compound than any other selection. 71 Figure 1A. -- Graphic presentation of the means for peak #7 for all the carrot selections included in the study. Figure 15. -- Graphic presentation of the means for peak #13, gamma-terpinene for all the carrot selections included in the study. 72 if 100% Full Scale = 36 7 UNKNOWN STD:61.1% PEAK A========_======__============= I I I P I I P on. 0000 0 0 a 9076 5432 100 ' s_xm..;.._t£15.__h_ a: 5.: 3:: B nu _===== o 1 1413 1385 1383 5987 .107 $11va SpFn 6de 5000 Gosin SELECTION 011111101 9. 3 l K A E P 100% Full Scale =4362 YJERPINENE $10137.4% 00 0 00 00 AU 0 09 876 54 3 z eat—wzeazzizai :2 i: 25» 107 SpSw SpFn (2de 6000 Cosin 1413 1385 1383 5987 SElECHON CARROT 73 Figure 16. —- Graphic presentation of the means for peak #21, isobornyl acetate for all the carrot selections included in the study. Figure 17. -— Graphic presentation of the means for peak #25 for all the carrot selections included in the study. 74 E “S .10. E: 8 0...... A3 9.. IL "MT. RI...“ 10 28%.... B an 000 $0.1 “II-ls A. r! B P B B .u___________—__.____.____________—___-.___.___—._______.__ no: 0 000 0000 0 76 5432 1. s...?:.: 315.5 s: 5.. 3:3 00 9 8 1413 1385 1383 5987 107 50va Soft: 6de 6000 Gosin —- SELECTION CARROT 3 1 2 IN a C Sun ""4 want 50F3 on u." N01. “"15 A E P A========_==.======_=========== M_======== 00 0 00 00 0 0 0 7.6 5432 9 8 2: x m2.q._:=z+z&.:u_ «:2 .2: 3:3 0 l 107 505' SpFn 6de 6000 Gosin 1385 1383 5987 1413 SELECTION CARROT 75 Peak Identification. Identification of peaks was facilitated by the use of mass spectrometry, literature references, and stand- ards, where available. In Table 1A, a listing of those peaks identified is given with reference to method of identification and confirmation. The compound indicated as peak #1 eluted on the capillary column at 7.80 minutes and on the packed column at 7.95 minutes. These are both very close to alpha- pinene's elution time of 7.82 minutes for the capillary and 7.96 minutes for the packed column. This alpha-pinene standard was 95% pure with a second peak at 9.00 minutes which constituted the other 5%, most probably in the form of beta-pinene. The mass spectral scan over peak #1 (Appendix K, Figure K1) was very similar to that reported by Buttery et al.(1968) and Ryhage and Von Sydow (1963) for alpha-pinene. The molecular ion was weak at 136 m/e. The first five major ions in decending order were: 93, 91, 92, 77, and 79 m/e. The M-A3 ion is 93 m/e and and is characteristic of the loss of an isopropyl group. Further transformation yeilds the 92 m/e and 91 m/e ions. The compound indicated as peak #2 eluted at 8.17 minutes for the capillary column and 9.A0 minutes for the packed column. The mass spectral scan over peak #2 is shown in Appendix K, Figure K2. The molecular ion is not intense enough to show up in this scan, however, the base peak is 121 m/e. Other researchers looking at total 76 Table 1A. —- Compounds identified through the use of Gas Chromatography and Mass Spectrometry of the volatile constituents trapped on the porous polymer, Tenax-CC. "52:13::"‘iiiicééézgéiiiiii33:53:77???2E2§i§§2§§§§i 1 alpha-Pinene + Buttery et.al.(1968) Ryhage/Sydow (1963) 2 Camphene " A beta-Pinene + n 5 Sabinene " 9 Myrecene + " 11 Limonene + " 13 gamma-Terpinene " 21 Isobornyl-acetate + 22 beta—Carophylene + 77 volatile components of carrots have indicated that camphene might be a likely candidate for this position in the chromatogram. camphene, like some of the other ter- penes, could very easily lose a methyl group from its structure during ionization in the mass spectrometer. This information, in conjunction with the base peak, might suggest a molecular ion = 136 m/e, which corresponds with the molecular ion of camphene as well as other terpinenes. The first five major ions found in the spectrum of peak #2 were: 121, 67, 95, 68, and 71 m/e. Ryhage and Von Sydow (1963) indicate a base peak of 93 m/e for camphene and Buttery et a1. (1968) support this finding. The major ions reported, other than the base peak, were 121, 79, and 67 m/e. The fact that the 93 m/e and 79 m/e ions were missing from the Spectrum of peak #2 indicates some question as to its being camphene. However, the remainder of the spectrum does look quite close to the rest of the reported spectrum for camphene. Eluting at 8.87 minutes on the capillary column was the compound indicated by peak #A. The same peak eluted at 10.95 minutes on the packed column. The elution times on both packed and capillary columns were similar to the 5% peak eluted during the injection of the 95% alpha- pinene standard. This second peak in the alpha-pinene standard was probably beta-pinene. Figure K3, Appendix K, is the mass Spectral scan of peak #A. The molecular ion was 136 m/e and the first five major ions were: 93, (A0, 78 A1,) 91, 69, 77, and 79 m/e. The ions A0 m/e and A1 m/e are placed in parentheses due to significant contributions to these peaks from the background noise. In referring to Ryhage and Von Sydow (1963), the strong ion at A1 m/e is typical of the beta—pinene Spectrum. Buttery et a1. (1968) makes no mention of this ion. The rest of the spectrum leaves no reasonable doubt as to its being beta- pinene since the pattern of ions are virtually identical in terms of the probability of intensity as that reported for beta-Pinene previously (Ryhage and Sydow, 1963). The compound indicated by peak #5 had an elution time of 9.53 minutes on the capillary column and 11.36 minutes on the packed column. Two mass spectral scans were run to either side of this peak to verify it's purity. Both scans were nearly identical and one is shown in Appendix K, Figure KA. The molecular ion was somewhat weak at 136 m/e and the first five major ions were: 93, 91, 77, 79, 136, and 9A m/e. The position of peak #5 in the chromatogram suggested Sabinene as the compound when compared to other chromatograms in the literature. Buttery et al (1968) reported a mass spectrum for sabinene that was very close to the findings reported here, and the published spectrum by Ryhage and Von Sydow (1963) was virtually identical. The compound indicated as peak #9 had an elution time of 11.13 minutes on the capillary column and 1A.0A minutes on the packed column. These retentions times were quite similar to those obtained for an injection of 79 myrcene. The retention times for myrcene were 11.23 minutes on the capillary column and 1A.09 minutes on the packed column. A mass spectrum scan of peak #9, shown in Appendix K, Figure K5, is very close to that of myrcene as indicated by Buttery et al.(1968) and also by Ryhage and Von Sydow (1963). The molecular ion was weak at 136 m/e and the first five major ions were 93, 69, A1, 91 and 77 m/e. The compound is very easily decomposed, as indicated by the very strong ions of 69 m/e and Al m/e. The retention time of the compound indicated as peak #11 was 11.88 minutes on the capillary column and 15.11 minutes on the packed column. These retention times were quite similar to those of limonene when injected on the same columns. The retention times for limonene was 12.08 minutes on the capillary column and 15.17 minutes on the packed column. In Appendix K, Figure K6, is shown the mass spectral scan of peak #11. The molecular ion was 136 m/e and the first five major ions were: 68, 67, 93, 79, and 9A m/e. This corresponds very nicely with the spectrum published by Ryhage and Von Sydow (1963) and closely matches the major ions listed by Buttery et a1. (1968) for limonene. The compound indicated as peak #13 had a retention time of 12.5A minutes on the capillary column and 18.03 minutes on the packed column. The relative position and time of retention in the chromatogram suggests that the peak could possibly be gamma-terpinene. A mass spectral 80 scan of peak #13 is shown in Appendix K, Figure K7. The molecular ion was 136 m/e and the first five major ions were: 93, 121, 136, 91, and 77 m/e. This mass spectrum agrees very closely with what Ryhage and Von Sydow (1963) published for gamma-terpinene. Buttery et a1. (1968) also published the identical list of major ions shown here for gamma-terpinene. The distinction between alpha-terpinene and gamma-terpinene is a very slight one in terms of mass spectral analysis. The distinguishing factor between the two is the base peak. Alpha-terpinene has a base peak of 121 m/e which is the molecular ion minus a methyl group (M-15) the sequence then follows 93 m/e (M-A3) and then 136 m/e (M). The mass spectrum for gamma-terpinene on the other hand has a base peak of 93 m/e which is the molecu- lar ion minus an isopropyl group (M-A3) the sequence then follows 121 m/e (M-15) and then 136 m/e (M). The differ- ence between the two Spectrum is caused by the different positions of the double bonds in the two compounds. The compounds indicated by peaks #21 and #22 have retention time of 17.1A minutes and 17.A5 minutes, respec- tively, on the capillary column. The same peaks have retention times of 25.08 minutes and 25.87 minutes, respectively, on the packed columns. These retention times are nearly identical to that of isobornyl acetate and beta-carOphylene on the same columns. Due to difficulties in background noise separation mass spectral data were not collected for these peaks. 81 Sensory Evaluation. Striver (1961) has calculated that as few as eight molecules of a powerful odorant are required for the triggering of one olfactory neuron in man and that as few as A0 molecules can produce an identifiable olfactory sensation. If an assumption is made that only one in 1000 molecules that are inspired ever reaches the olfactory -19 region, then A0,000 molecules or about 10 moles can, at least theoretically, be detected by the nose. Scientist continue to investigate those odorants within the range of our analytical techniques and utilize methods of aroma enrichment as with the porous polymer trapping technique to produce quantities within range of the analytical techniques. Few scientists take the time to realize the true complexity of our olfactory system. The Stimulus-Response Circuit diagram in Figure 18 is a breakdown of the Signal path for a nerve network. As Shown, the raw (true) stimulus is received by the receptor. The receptor passes it's "reading" on to the processing of stimulus section of the central nervous system by way of the afferent nerves. The brain retrieves the processed stimulus message and verifies it. The brain can then evaluate and pass judge- ment on the stimulus, interpeting it based upon it's inherent information and application of logic (or ill- 1ogic) and with a multitude of inherent biases possibly 82 :36 a.“ vasou fined“??— ufisoufio wmaonmwm I maaaafium 93 mo ment-z 5:: .0 9.730005 2.32.5 .0 9.76000... wm>mw2 11 hzmcmuum 1 mmzmz hzmmmuu< uuumaosom II .wH munwwh EOhOwuuw cOhamme 1 02083. It All m:_:EZm II 83 being applied. At this point a proper response is formu- lated and transferred to a Process of Action section in the central nervous system from the brain. Efferent nerves then carry a message to the effectors which produce the response. There are many places in the circuit where variations can appear and differences manifest themselves. The psychology of the process is very involved and it is of utmost concern that the researcher be very aware of this. A. Modified Open Discussion Profile Panel. The modified open discussion panel was designed to not only collect commonly agreed upon and discussed de- scriptors but also any descriptors not discussed yet commonly noted on the ballots of 50% of the judges. The histogram in Figure 19 shows the various aroma character- istics listed by the judges as possible descriptors for the aroma of raw carrots. Of the total number of descriptors listed, 62% were accepted as viable. Those descriptors that were rejected were: stale, rancid, green, bitter, aromatic, and pungent. It is interesting to note that the "green" descriptor was eliminated from the listing. This descriptor has been implicated in the "green toppy" notes of the raw carrot (Buttery et al., 1968; Heatherbell et al.,l971; Alabran et al., 1975), especially in reference to the carrot greens. One might suspect that in this case, due to the large number of 84 .Hoch :owmwsomfip ammo msu Eouw vmuwuocow mofiumfiuwuomumsu meoum muomcmm wow Emuwoumwm II .aH wusmam .............. ... ................................... .... ... .... ... .... ... .... ... .... ... .... ... .... ........... .... ... .... ... .. AmzzOmZm—m AONBHOEHd 85 descriptors, that other descriptors might adequately explain the aroma characteristic described as "green". Heatherbell et a1. (1971) tied the "green" aroma to "earthy" calling it a "strong green-earthy" aroma and the earthy descriptor was included in the listing chosen by the open discussion panel. Martens et al. (1979) tied the "green" aroma to "grass" calling it a "green grass" aroma. This might be similar to the hay-like and/or piney descriptors. The remaining descriptors were: sweet, carroty, perfumy, potato-like, woody, musty, hay-like, piney, and earthy. Of these descriptors, all but one were used in the testing-training phase and the Qualitative Descriptive Analysis. The one descriptor removed was potato-like. This was removed by the analyst because of the existence of reasonable doubt surrounding the validity of this descriptor. The reasonable doubt is based on the fact that just prior to going into storage, other storage chambers located off the same corridor were filled with freshly harvested potatoes. Any exchange of air in the storage cubicles took place between the corridor and the cubicles giving ample opportunity for exchange of volatiles. In this particular wing of the controlled environment facility, six out of the 13 chambers were used for potato storage. One was used for the storage of carrots. On the basis of these conditions the descriptor, potato-like, was removed from the listing of valid descriptors of raw carrot aroma. 86 In the second portion of the open discussion panel each judge was asked to rank the two sets of carrot selec- tions in order of preference. The results were analyzed in two ways: (1) using Analysis of Variance and the W coefficient (Kendall, 19A8) and (2) Kramer's Rank Sum Method (Kramer, 1960). The coefficient of concordance (W) applied to the Analysis of Variance was used to determine if there was significant agreement between rankings made by the judges. No significant agreement was found for either group of samples using the coeficient of concordance method (Appendix L). The Kramer's Rank Sum Method was used to determine if differences existed between samples. Of the first set of five samples, no difference was noted between samples. Of the second set of samples, one was determined to be the worst. The exercise of requesting judges in the open discussion panel to rank samples was simply to evaluate the ability of the judges to rank similarly some quite different samples. The results might suggest that the judges disagree on what constitutes the best raw carrot aroma. This is a desirable attribute in the Open discussion panel. Differ— ences can be discussed and opinions can be shared. The differing attitudes can be molded into a reasonable consortium of descriptors capable of embodying many varied opinions. The fact that the judges appeared to disagree on the ranking of the samples also supports the modified method of open discussion panel used in this study. The 87 possibility exists that certain judges may unduly over- influence the discussion. With the modified method, some degree of protection is applied to the situation. B. Testing-Training Panels. Dawson and Harris (1951) stated: "Successful con- duct of taste panels is frequently as much a matter of human relations as a scientific problem. Panel members must have a keen interest in their tasting ability and these feelings must be sustained." Success breeds suc- cess, may be a common phrase, but how often is it neglect- ed in terms of sensory panelist's performance. Just as valid might be the statement, failure breeds failure. Success develops attitudes of self-confidence and a desire to succeed. These concepts can well be applied to the sensory evaluation studies and for optimization, should be included in the development of procedures designed to promote the panelist, into a mode of achievement for success. Success appears to be dependent on the desire to excel per se and the desire to do better than other sub- jects. Henderson and Vaisey (1970) found that judges selected on the basis of high scores in need for achieve- ment performed better than low scorers in flavor difference test. They further noted that throughout the test period high achievers showed somewhat better discrimination of moderately difficult comparisons. 88 Of the 37 original panelists, 5 dropped out of the study, 1A failed to pass the required tests as outlined in the methods section and 18 panelists completed the testing training phase and continued on to the sensory evaluation of actual samples using the Qualitative Descriptive Analysis method. Motivation of panelists in the raw carrot aroma study was incorporated into the design of the project through the use of devices that developed a sense of achievement, especially for those panelists who passed the testing-training phase and continued on into the Qualitative Descriptive Analysis part of the study. One of these devices (Appendix H, Figure H1) consisted of personalized ballots. The ballots were generated by a computer data base maintained on each of the panelists. The ballots included direct addressing of the panelists and information concerning previous achievements in tests. A newsletter (Appendix M, Figure M1) was generated to keep the participants of the panels in tune with the purpose of the study and the various stages of the study. For those participants making it through this phase of the study, a "diploma" and "union membership card" was generated for each, again to reinforce the motivational factors behind a sense of achievement. 89 C. Qualitative Descriptive Analysis. In this part of the study two measurements were made for each aroma descriptor (Figure 20.). The measure— ment for intensity was simply a quantity measurement for the characteristic of interest. The second measurement was more of a subjective measurement as to how the panelist view the previously measured quantity. The second measurement was labeled as Desirablility. A one way analysis of variance was applied to the Qualitative Data Analysis to investigate the possibility of individual sensory descriptors varying between carrot selections at a statistically Significant level. Mean square values and degrees of freedom for each are shown in Table 15. AS noted, the panelists were not able to give descriptive profiles of the raw carrot aroma that varied at a significant level between carrot selections. A summary of the results for the Qualitative Descriptive Analysis study is presented in Figure 21. This figure is a model for the aroma attributes of carrots in general, based upon the mean values for each descriptor. The solid line describes the ratings of intensity. The circle indicates the a value of five on the 0 to 10 scale. The dotted line describes the desirability rating for each descriptor. In general the carrots included in this study had close to a medium level of plain raw carrot aroma and this level was slightly below what is typically believed by the panelists to be optimum. The piney character was somewhat low but considered close 90 .mfimmamc< o>wuafiuumma 0>wumuqawso ecu mo xquESm II .ON wusmfim <20m<|44<¢w>0 mx77>hm35 >000; >Ih¢22umma >m2:_ >h.=¢u >h0mz<0 mQthflommeo mEo.< #10:.bm33 020...; 00» 0255 ooh :9: 22.3.2 So. 8 a .... n m S ( >t.:m<.:mmo >tmzm..z_ Eco. m_m>.._.E_mommo m>:<:._<:o 91 Table 15. -— Mean Squares and degrees of freedom as analyzed using One Way Analysis of Variance for the Qualitative Descriptive Analysis Data. Analysis of Variance for Sensory Data Variable Between Groups Within Groups "STE-"Kg" "STE-"Ti; Q01 - Carroty Intensity 9 7.0A6A 3A7 5.AA18 Q02 - Carroty Desirability 9 3.3779 3A7 3.01A6 Q03 - Piney Intensity 9 6.A668 3A7 5.7363 QOA - Piney Desirability 9 1.6363 3A7 1.1568 Q05 - Sweet Intensity 9 7.0557 3A7 A.1812 Q06 — Sweet Desirability 9 2.8AA8 3A7 2.1366 Q07 - Woody Intensity 9 A.8688 3A7 6.9132 Q08 - Woody Desirability 9 1.2687 3A7 1.5A63 Q09 - Hay—like Intensity 9 5.2003 3A7 7.5A53 Q10 - Hay-like Desirability 9 3.1A66 3A7 1.9935 Qll - Fruity Intensity 9 3.A268 3A7 3.3576 Q12 - Fruity Desirability 9 .9029 3A7 2.0025 Q13 - Perfumy Intensity 9 3.757A 3A7 A.0739 QlA - Perfumy Desirability 9 1.2802 3A7 1.3AA5 Q15 - Earthy Intensity 9 1.2029 3A7 7.9795 Q16 - Earthy Desirability 9 1.AA63 3A7 1.A03A Q17 - Musty Intensity 9 .8298 3A7 7.1871 Q18 - Musty Desirability 9 1.10A6 3A7 1.2621 Q19 - Overall Intensity 9 A.689A 3A7 3.7A98 Q20 - Overall Desirability 9 1.5A16 3A7 2.A52A 92 .ucuauuomvt :uzm no. mwzqm> :cwE com: twang muouuco uo mauzcduuun Asomcom use #0605 mquanc< o>fiuafiuommc m>qumufifimsc II .HN Shaw“: 0— o a s o r. v n a u 0 P P p — p P _ n _ p — 55E >225; 53.2: \\\.. 21.x, s i llllllIttvloll \ >oOO>> . >IE>m ... \ >522 . . o /. >w7=m .111.. .<<§Umd. :<~E>O 50520 2.82:. ---- £5258 93 to optimum levels. The sweet aroma in the raw carrot head Space was quite low and appeared to be lower than desired by the panelists. The woody aroma was at a similar level of intensity as the sweet aroma but was considered to be close to optimum. This same statement holds true for hay— .1ike, earthy, and musty. The fruity and perfumy aroma characteristics were both rated very low for these carrot selections and yet considered just below an optimum level. The panelists were also required to place an overall rating on the level and desirability of the odorants from the raw carrot. The overall level of intensity was medium and this level was slightly lower than desired. Factor Analysis. Factor-analytical techniques enable us to see if some underlying pattern of relationships exist so that data may be rearranged or reduced to a smaller set of components or factors that may be taken as source variables accounting for the observed interrelations in the data. The study of aroma characteristics is very complex and poses an ideal application of factor analysis. The variability built into the study ranges from that associated with the raw carrot itself to the complexity of the psychological differences among panelists. To some degree there is an attempt to reduce or "commonize" this variability found on the psychological level through the 9A use of training sessions and/or screening sessions, however, seldom can this goal be achieved completely. In this study, psychological biases are heightened by the use of a deliberate measurement of opinion, the question of desirability for each aroma characteristic. With the intention of using the newly calculated factor variables in multiple regression analysis the question of lost mean- ingful variation is sometimes raised. Rummel states in his overview of the applications of factor analysis that the composit variables may be used in the regression analysis in place of the original variables with the knowledge that the meaningful variation in the original data has not been lost (Rummel, 1967). Data or facts are meaningless in and of themselves. It is only when we apply theory to them that we approach the aim of science. Once linked through propositions, an interpretation or meaning can be conferred upon the assem- bled data. This is the role of factor analytical tech- niques, that is, to expose and determine these linkages and define them so that the researcher might interpret them. The factor model represents a mathematical formal- ism departing from the calculus functions of classical physics. The analytic part of the factor model, that part involved in the separation of the whole into the component parts, is akin to that of quantum theory. An understanding of the patterns defined by factor 95 analysis can be enhanced through a geometric interpreta- tion. Each of the carrot selections in this study can be thought of as defining a coordinate axis of a geometric space. Although a pictorial presentation of this model is limited to three dimensions; the Space defined above would have a total of 10 dimensions, one for each of the carrot selections included in the study. In the space defined by the 10 dimensions each sensory characteristic can be considered a point located according to its value for each carrot selection. For each point a line can be drawn from the origin to that point for a vector presentation of the data. The twenty vectors, each representating a sensory characteristic variables (10 related to Intensity measurements of the descriptors and 10 related to Desirability measurements), would then describe a vector space. The angle between any two of these vectors is a measure of the relationship between the two sensory characteristics for the ten carrot selections. The closer the angle to 90 degrees the less the relationship. The closer the angle to zero degrees the stronger the relationship. Obtuse angles indicate a nega- tive relationship and at the extreme, an angle of 180 degrees between two vectors means the two characteristics are inversely related. The cosine of the angle between vectors is, with minor qualifications, equal to the pro— duct moment correlation coefficient between the sensory characteristics represented by the vectors. 96 With the vectors describing a Space, their configuration then reflects the data interrelationships. Sensory characteristics that are highly interrelated will cluster together; characteristics that are unrelated will be closer to right angles. Any clusters that are found, index patterns of relationships in the data: each cluster is a pattern. What factor analysis does geometrically is enable the clusters of vectors to be defined when the number of cases (dimensions i.e. carrot selections) exceeds our graphical limit of three. As a result, each factor delineated by factor analysis defines a distinct cluster of vectors. Factor analysis mathematically lays out a vector space and then projects an axis through each cluster. This is analogous to giving each vector one unit of mass and then allowing the center of gravity to define factor axes. The projection of each vector, in this case a sensory characteristic, on the factor axes defines the clusters. The projections are called loadings and the factor axes are termed factors or dimensions. The algebraic factor model can be described by the following equation: Y . a F + a F + ... + a F n n1 1 n2 2 nm m where: Y . a variable (sensory characteristic) with known data. a = a constant (factor loadings). F 8 a function, f( ), of an unknown variable. ( A factor or dimension) 97 The F stands for a function of variables and not a true variable. The unknown variables entering into each func- tion, F, are related in unknown ways, although the equa- tions themselves are linear. From the application of factor analysis the unknown functions are defined. The loadings calculated from the analysis are the "a" constants. The factors are the F functions and the size of each loading for each factor measures how much that specific function (factor) is related to Y (a measured variable). The sensory evaluation data were submitted for analysis using factor analytical techniques. The correla- tion matrix is shown in Table N1 of Appendix N. In this 20 by 20 matrix, coefficients of correlation express the degree of linear relationship between the row and column variables of the matrix. The principal diagonal and the upper half of the matrix has been deleted for conciseness of presentation. The principal diagonal consisted of all one's and the upper half was the mirror image of the lower half which is Shown. Initial factoring, using Alpha Factoring, produced the Unrotated Factor Table shown in Table N2 of Appendix N. In Alpha Factoring, variables included in the factor analysis, are considered a sample from the universe of variables. In the Unrotated Factor Table, the columns define the factors, the rows pertain to the variables, and each intersection of row and column is the loading for the 98 particular row variable on the column factor. Five dis- tinct patterns (factors) are observed in the sensory evaluation data. The first factor accounts for the great- est regularity in the data and each successive factor has been fitted to best determine the remaining regularity. At this point, the patterns in the data have been account- ed for but the distinction of clusters have not. This is why the application of factor rotation is commonly the next step. The rotation of factors was performed using the Varimax criteria system which strives to maximize and minimize loadings for ease of understanding and interpre- tation. The Varimax criteria system does maintain orthog- onality such that factors defined, as independent through alpha factoring and then rotated using Varimax are main- tained orthogonal. This is important as mentioned before for application of multiple regression analysis, the final step in this study. In Table 16, the variation characteristics of each factor are Shown. Eignvalue is a method of expressing the amount of variation accounted for by a factor. As can be seen from the table, the factors are ordered in terms of decreasing accounted variation. The rotated factors are Shown in Table 17. The odd numbered "Q" variables relate to the respective aroma intensity and the following even numbered "Q" variables relate to the respective aroma desirability. 99 Table 16. —- Summary of data variation explained in the factoring of the sensory evaluation. Factor Eigenvalue % Variance Cum. % Cum. % (based on Tot.) (Total) (based on 5) Factors 1 5.22118 26.1 26.1 A0.5 2 3.6A2A6 18.2 AA.3 67.9 3 2.11A67 10.6 5A.9 83.2 A 1.67A55 8.A 63.3 93.2 5 1.19096 6.0 69.2 100.0 Table 17. -- The rotated factors for the factor analysis of the sensory evaluation data using the Varimax Criteria system of rotation. Var. Fact(1) Fact(2) Fact(3) Fact(A) Fact(5) Q01 -.10991 .8A602 .06197 -.0A853 .006A8 Q02 -.08A80 .7878A -.0AA81 -.0352A .02829 Q03 .136AA -.00018 .33630 -.2598A .620A1 QOA .2857A .11693 .06910 .01790 .56863 Q05 -.101A3 .AA588 .A56AA .165A1 -.11116 006 -.17622 .5655? .0053A .A1520 -.17382 Q07 .5379A -.10208 .25962 -.36007 .30383 Q08 .67103 .00321 -.0561A .01811 .30605 009 .57868 -.07092 .36A90 -.35559 .19665 Q10 .76021 .00A57 -.0279A .03288 .28718 Q11 -.03311 .lA28A .6A538 .03098 .16538 Q12 -.15201 .39105 -.08AA8 .51985 .02521 013 .06778 -.11356 .75381 .05756 .1321A QlA .1A861 -.08968 .22300 .83278 -.13519 Q15 .50830 -.09897 .A6912 -.5AA86 -.03198 Q16 .6AA61 .0A959 -.01158 -.11956 .02758 Q17 .A9159 “007832 060691 “036695 -000796 Q18 .70050 .03160 .09833 .07307 -.0A370 Q19 .25791 .73363 .0A909 .10A60 .12A32 Q20 .38569 .65A66 -.0505A .03810 .1AA75 100 Figure 22 is a summary of the interpretation of this factor analysis. Factor 1, accounting for a majority of the explainable variation is characterized by the term: "Earthy - Organic Aroma". It included both intensity and desirability characteristic for the sensory attributes measured with the descriptors: Woody, Hay-like, Earthy, and Musty. These were all, including both the intensity and desirability questions for each, loading on this first factor positively. It is somewhat unexpected that this factor would account for more variation than that of the second factor, termed: "Basic Raw Carrot Aroma". This second factor consisted of the following aroma variables: Carroty, Sweet, and Overall. These were also positively loading on this second factor including both intensity and desirability questions for each. The third factor, for lack of a better name was termed: "Intensity of Aromatics Other Than Carrot". This describes a measured variation defined by the intensity rating of the following aroma variables: Fruity, Perfumy, Musty, Earthy, and Sweet. Although the Sweet aroma is connected to the Basic Raw Carrot Aroma it is just slightly within the correlation indices for the acceptance range on this third factor. The fourth factor is termed: "Desirability of Pleasant Aromatics (Non-Earthy)". The desirability ratings for Sweet, Fruity, and Perfumey load in a positive direction on this factor and the intensity rating of the 101 .mumv cowumaam>w muomcom msu cu vmaaaam mamzamcm wouumw ecu mo :ofiumuwnuuucfi 0:» ac mumasam II .NN madman 6.. sec... Rama 806 sz "0:32 I m m0h0E=toa\>=:.u\.ooim 25.31.02. moi<20m< hzE=..oa\>=:.u\.oozm #0255 ....o mo...<20¢< .0 >.:mzwhz. H25.2 a a £9.05. A... 9:03 ..a.o>0\.oo3m\>.o:oo $6.3. and. $20.2 bongo 22: 0.35 85oz - a coho: .o._ ...3.21.5.3).573:}..33 :50 8—6N 8—.o~ <20m< 92(010 >Ihc 8.: All mc0h0mOmZm—m ...O m_m>._<2< $05.04*“. 102 Earthy aroma loads negatively on this factor. The last factor explaining a portion of the variance is termed: "Piney Aroma". It is the only individual aroma variable that was determined to be completely independent, defining it's own factor. With the first stage of the factor analysis complete, the sensory evaluation data has been interpreted into five definable and completely orthogonal characteris- tics. These characteristics were then taken as factor variables and calculated on a basis of the equations in Table 18 for all of the sensory evaluation data base. Table 18. -- Equations used for the calculation of new Factor Variables VBl = (.5379“ x Q07) + (.67103 x Q08) + (.57868 x Q09) + (.76021 x Q10) + (.50830 x Q15) + (.6NN61 x Q16) + (.N9159 x Q17) + (.70050 x Q18) VB2 = (.8u602 x Q01) + (.7878u x Q02) + (.uusas x Q05) + (.5655? x Q06) + (.73363 x Q19) + (.65N66 x Q20) VB3 = (.useuu x Q05) + (.6N538 x Q11) + (.75381 x Q13) + (.h6912 x Q15) + (.60691 x Q17) VBM = (.u1520 x Q06) + (.51985 x Q12) + (.83278 x Q1“) + (-.suuae x Q15) VBS = (.62041 x Q03) + (.56863 x 00“) 103 The equations express the factor variables (mnemonics: VBl, VB2, VB3, VBM and VBS) in terms of each significant sensory variable with loadings used as coefficients for the variables (Anderson, 1980). A second factor analysis was performed on the newly calculated factor variables along with the peak area data in order to determine which peaks, if any, should be tested for prediction of factor variables in multiple regression analysis. In Table N3, Appendix N, is shown the variation accounted for by the nine factors determined for this second factor analysis. The factor matrix was rotated using Varimax criteria. The resultant matrix is shown in Table NA, Appendix N. The first three factors involved only peak area data and thus would not be included in multiple regression analysis. The range for acceptance for inclusion in the multiple regression analysis was extended to the regions of .2 to,1 and -.2 to -1. This was to ensure that even the variables loading quite low on a factor be tested for possible significance in the regression equation. The fourth factor included four of the factor variables. Factor variables VBl - Earthy aroma; VB3 - Aromatics O.T. Carrot ; VBS - Piney Aroma all loaded quite highly on this factor and VBH - Pleasant aromas, loaded quite highly but negatively. There were three peaks very weakly associated with this fourth factor: P15, unknown; P16, unknown and P23, unknown . All 10h three peaks varied very little, if at all, between carrot selections and P15 and P16 only showed up in MSU-lul3 and MSU-6000 at very low levels. Factor five had two factor variables loading marginally, VB2 - Basic Raw Carrot Aroma and VBS - Piney Aroma as well as seven peaks including camphene, beta-pinene, sabinene, limonene, and two more unknowns P1“ and P2A. This is especially interesting in that the Basic Raw Carrot Aroma is slightly associated with compounds classically found in all carrot extractions and volatile analysis. Factor six includes only peak area data. Factor seven had factor variables, VB2 - Basic Raw Carrot Aroma and VBM - Pleasant Aromatics loading on it with VB2 loading quite strongly. Additionally three peak variables appeared to be associated with the factor, P15 and P16 both unknown and loading negatively, and P17 - unknown, marginally positive. Factor eight involved only peak area data. Factor nine had factor variable VBS - Piney Aroma and six peaks loading on it, camphene and beta-carophylene loading positively, and unknown peaks P03, P17, P20, and P2A, loading negatively. The peaks associated with the factor variables were submitted for multiple regression analysis. The predic- tion equations tested for significance are in Table 19. Of the five equations tested only the equation predicting factor variable VBl — Earthy Organic Aroma was signifi— cant, however, none of the variables in that equation could be shown to be significant. 105 Table 19. - Multiple Regression Analysis of prediction equations for each factor variable involved in the second factor analysis. Factor Peak Variable Constant Variable VB1* = 21.929 + 0.u98 (P15) + 0.1h5 (P16) - 0.0A7 (P2A) VB2 = 16.08u + 0.113 (P02) + 0.180 (P15) + 0.893 (P17) + 0.0007 (P2A) + 0.057 (PIA) + 0.027 (P11) - 0.001 (POA) — 0.A65 (P16) - 0.1u0 (P03) VB3 = 7.656 + 0.0930 (P15) + 0.0901 (P16) - 0.02A1 (P2A) VBu = 5.uu + 0.013 (P15) + 0.273 (P17) + 0.012 (P2A) - 0.173 (P16) VB5 = 5.121 + 0.009 (P02) + 0.09A (P15) + 0.0uu (P20) + 0.103 (Pin) + 0.0008 (P11) + 0.008 (P16) + 0.155 (P22) - 0.133 (P17) - 0.013 (P2A) - 0.0002 (P05) - 0.001 (Pou) - 0.113 (P03) 5 Significant at the 95% level 106 Computer Software for Sensory Analysis. The use of computers for direct input of data is coming of age. With the sharp decline in cost over the past ten years and the new availability of micro- computers, the computer as a data logging tool is becoming more and more prevalent. In this study we have experimented with direct interfacing of the sensory panelists to the computer via the use of a crt (cathod ray tube) terminal. Special care had to be taken to format an approach that would be simple and easy to under— stand, yet accomplish the measurement of some sophisticated parameters. The approach decided upon was to emulate the ballot method of input, which most panelists had previously used for sensory evaluation. The final program is shown in its entirety in Appendix J. The program establishes the conditions of the panel and then enters a mode of self- initiation. Lines 1000 to 1680 in the computer program are where the computer is informed of the panel condi- tions. Room is cleared for operation in line 1190 and the computer is told wether there is an old file in existence to which the following panel information should be added. The section in lines 1300 to 1&00 instruct the computer to read in an old data file if in existence and to re-write the file in ASCII (American Standard Code for Information Interchange) format to a sequential output file. Sequential files were used for storage because of the 107 convenience of transfer between the TBS-80 microcomputer and the Control Data Computer used for statistical analysis. The sequential files are written to disk in an unabbreviated ASCII format where as a Direct Access file is written in a condensed, abbreviated form. The ASCII format permits the direct transfer of the sequential files to the Control Data Computer without an interpretation in between. This type of application for microcomputers is very popular i.e. working as a mini- "front end computer" to a large "main frame" computer. Following the reading of old data, the program instructs the computer to load variables with the panelists names, the panel number and the number of samples. Lines 16N0 to 1690 describe the instruction for entering the sample number and associated random number in the computer. At this point the computer set up is complete. The computer now waits for a panelist to enter their last name . Upon entry of a name, the computer then randomizes the order of the samples and then checks to make sure that the panelist name is valid. From here, the panelist is personally welcomed and told which of the randomly ordered samples is his/her first one to sniff. The video screen constantly updates the panelist as to which sample he is on and what question he is answering. The qualitative analysis ballot usually involves placing marks on a linear unsegmented scale associated with the sensory characteristic of interest. This is accomplished 108 through the use of a light bar on the screen. The panelists move an arrow to the appropriate location to place an "X" on the bar. Once the "X" is placed, the computer measures the distance to the "X" and logs this value under the coding for the question number, panelist, sample number, and panel number. Both the intensity and the desirability questions show on the screen together, however, only one aroma characteristic is shown on the screen at one time. Panelists have an option to re-do the reponses given for a particular aroma immediatly after completeing the two questions of intensity and desirabili- ty. This operation continues until all twenty responses are collected for the sample. At this point the coded data is then transferred to permanant storage on the floppy disk system to be later transferred to the Control Data Computer for analysis. At the end the panel when all panelists have finished the analyst enters the name "END" then the computer organizes itself , closing open files and then shuts down. Care was taken to ensure that each panelist under- stood exactly how the computer would escort them through the questions. Computer interaction with the panelists was made as personable as possible through the use of a data base which could be called upon by the computer to find out the panelists first name. Total estimated time saved on the part of the analyist through the use of the computer is approxiamtely 109 eight hours. This figure is based upon a trial analysis set up for the purpose of testing the amount of time saved. Each ballot took five minutes to measure and code. Assuming a minium of three samples per panelist per panel this leads to 15 minutes per panelist. If two minutes are added for paper shuffling and rest between every three ballots the total comes to 17 minutes. This times the 18 panelists yields Just over five hours. Another three hours could easily be added for the keypunching time. Thus a minimum of eight hours labor per panel of this magnitude can be saved. CONCLUSIONS The use of the Tenax-GC as the packing material for the porous polymer trap has postive, as well as negative aspects. The porous polymer afforded an opportunity to collect and concentrate headspace volatiles without many problems commonly associated with the classical methods of headspace analysis. The volume of polymer required in the trap to efficiently collect compounds was very little, in this case 0.01g per trap. Trap regeneration was very efficient, with the passage of approximatly 1 ml of ethyl ether sufficient for 100% regeneration. One of the most appreciated aspects of the porous polymer trap techniques is that artifacts are nearly eliminated with the retention of volatile ratios close to that found in the original food stuff. The porous polymer, Tenax-GO, exhibited high retention time characteristics for the higher boiling compounds with more care required for the collection and concentration of the lower boiling compounds. The drying procedure had to be severely shortened in order to compen- sate for the possible loss of volatiles. Although the Tenax-0C is reported to be the best polymer in terms of 111 retaining volatile compositions close to the state of the original food stuff; there might be a possibility of combining some of the newly developed polymers with the Tenax-CC in order to achieve optimum trap characteristics where low and high boiling compounds would both, be highly retained in ratios similar to that found in the original food. The modified open discussion panel appeared to perform as expected. Assuming a null hypothesis that all descriptors are alike and not siginificant; the concept behind the modification is akin to protecting against Type II Error in statistics, that is the error of rejecting a descriptor as being invalid when in fact it is a valid descriptor. Although the application of the concept heightens the probability of Type I Error, the error of accepting a descriptor as being valid, when in fact it is not valid; the later application of factor analysis should more than compensate for this by ensuring exposure of indifferent descriptors and formulation of new orthogonal descriptors. During the testing—training phase of the study, an attempt was made to heighten the sense of achievement of the panelists. It was felt, based on feed back from the panelists, that the achievement factor was entering into how well the panelists were performing the Job of identifying classes of aromas in this phase. Seeing their previous scores, panelist commented that they would 112 attempt to top a previous score or record of high scores. Although the presentation to the panelists of a diploma and union card was taken very light heartly; panelists commented on the sense of belonging and of specialness of the panel. In the qualitative descriptive analysis study panelists were not able to describe differences between carrot selections. This condition may have arisen from numerous factors. The panelists may well have been able to distinguish differences in the aromas but when requested to elucidate those differences by way of descriptive ratings, they could not. Another possibility is that the training phase of the study did not achieve the uniformity of understandings and capabilities hoped for. The application of factor analysis to sensory analysis is not very common. In this application I was able to define from the sensory analysis a new set of interpretations of the same data consisting of five new factor variables. The five sensory descriptors were (1) Earthy organic aroma, (2) Basic raw carrot aroma, (3) Intensity of aromatics other than carrot, (A) Desirability of pleasant aromatics (non-earthy) and (5) Piney aroma. The fact that factor analysis has the capability to achieve the recalculation of new factor variables without loss of the original data is a very powerful tool. In this application five new variables, completely 113 independent of one another were established for use in a second application of factor analysis. The second application of factor analysis allowed us to formulate possible relationships which could be tested with the use of regression techniques. Each of the new factor variables, identified as sensory parameters, had various peaks associated with it from the headspace analysis which were tested in a multiple regression analysis. Of the sensory characteristics, only one equation indicated significance and that was VBl - The Earthy Organic Aroma in carrots with peaks 15, 16, 2“, each at trace levels, too low to identify. Although the regression trend was significant when each variable (compound) in the equation was tested, none were signifi- cant. The possibility exists that the regression and variables are significant and that in this case variations in the peak areas for the compounds were too high to support the findings, yet low enough to indicate a signi- ficant trend. The results of the study tend to support the belief that the taste parameters are far more important in the acceptance of a "good" carrot than the aroma characteris- tics. The implications to the carrot breeder are straight forward; breed for a better tasting carrot and generally disregard minor variations in associated aroma characteris- tics. The use of the computer as a data gathering tool 11“ seemed to generate vast amounts of interest by the panelists. It was felt that in this study, the computer eased the laborious process of requesting feedback for some twenty questions for each sample. The application of computers in the field of sensory data gathering has a potential for growth. The time savings can be recognized and interpreted into a cash savings due to the lowering cost of hardware investments. This concept may have far reaching effects in terms of industry applications. Further research appears to be unbounded in terms of direction and depth. An interesting area of study may be the use of micro-computers to not only collect response data from a panelist but also to monitor bodily changes as secondary or subconscience response to stimuli applied to the subject. APPENDICES Appendix A. Volatile Constituents of Carrots Table A1. Volatile Constituents of carrots identified by Buttery et al., 1968, 1978, 1979. 1968. -- Buttery et al. alpha-pinene camphene sabinene beta-pinene myrcene alpha-terpinene p-cymene limonene gamma—terpinene terpinolene caryophyllene beta—bisabolene gamma-bisabolene 1978. -- Seifert and Buttery alpha-bergamotene beta-farnesene alpha-humulene gamma—muurolene gamma-bisabolene-A * gamma-bisabolene-B * * - originally thought to be beta and gamma now believed to be isomers noted as A & B. 1979. -- Buttery et a1. geranyl 2-methyl-butyrate geranyl isobutyrate beta-ionone geranylacetone p-cymen-8-ol elemicin eugenol p-vinylguaiacol h-methylisopropenylbenzene 116 Appendix A. (cont.) Volatile Constituents of Carrots Table A2. -- The volatile constituents of carrots identified by Heatherbell et al., 1971, 1971, 1971. 1971. -- Heatherbell et a1. diethyl ether acetaldehyde acetone propanal methanol ethanol alpha-pinene camphene beta-pinene sabinene myrcene alpha-phellandrene limonene gamma-terpinene p-cymene terpinolene octanal unknown '1 H " 2-decenal unknown " bornyl acetate caryophyllene terpinene-A-ol sesquiterpene-a beta-bisabolene gamma-bisabolene unkown sesquiterpene-b sesquiterpene-c unkown sesquiterpene-d carotol unknown myristicin 1971. -- Heatherbell and Wrolstad, 1971a. - same as above - Table A3. Table AA. Table A5. 117 Appendix A. (cont.) Volatile Constituents of Carrots 1971. -- Heatherbell and Wrolstad, 1971b. - same as above - Volatile constituents in carrots identified by Murray and Whitfield, 1975. 3-isopropy1-2-methoxypyrazine 3-sec-butyl-2-methoxypyrazine Volatile constituents in carrots identified by Cronin and Stanton, 1975. 3-sec-butyl-2-methoxypyrazine ** ** - claimed as very important odorant in carrots. Volatile constituents in carrots identified by Linko et al., 1978. formaldehyde acetaldehyde acetone propanal 2-methy1propanal Butan-2-one n-Butanal 3-hydroxy-2-butanone 3—methylbutanal pentan-2-one buten-2-a1 n-pentanal methylbutanal n-hexanal n-heptanal 6-methyl-5-hepten-2-one 5-methylfurfura1 n-octanal n-nonanal unknown decen-2-al u-Undecanal n-dodecanal alpha-ionone 118 Appendix A.(cont.) Volatile Constituents of Carrots Table A6. -- Volatile constituents in carrots identified by Simon et al., 1980. alpha-pinene beta-pinene sabinene myrcene alpha-phellandrene alpha—terpinene limonene gamma-terpinene terpinolene terpinen-h-ol bornyl acetate caryophyllene gamma-bisabolene (a) gamma-bisabolene (b) 119 Appendix B. Relative Humidity Contol System 110 AC 24 Hour Timer 1 Host Switch ——\ I I! \ ----"—E—Ji ! S i 5 fl 5 SSgaHon 3 Distilied Water 1 Pump ‘ \ J Figure 81. -- Schematic of the control system for maintanence of relative humidity in the environmental storage chamber. 120 Appendix C. Volatile Collection System Tenax-6C 1}ap Sample Vessel Figure C1. -- Schematic of the volatile collection system using the porous polymer traps and nitrogen sweep technique. 121 Appendix D. Porous Polymer Trap Elution System —T— iw’r—‘J Scc Syringe Plastic Manifold \ Porous Polymer Trap Sample Holding Vessel Figure D1. -- Schematic of the trap elution mechanism. 122 Appendix E. Gas Chromatography Void Volume, Flow Rate and Split Flow Calculations Column Length = 25 meters Column diameter = 0.2 mm. 2 pi x r x h where pi = 3.1h r = radius h = height or length Total Volume 2 Total Volume = 3.1M x (0.1 mm.) x (25 m.x 1000 mm./ m.) = 785.3 cu. mm. 785.3 cu. mm. / 1000 cu. mm. per cc. 0.785 cc. An injection of Methane a non-retained compound gave a retention time of 1.50 min. Flow Rate = 0.785 cc. / 1.50 min. = 0.52 cc. / min. Split Vent Flow = 60 cc. / min. Split Vent Flow Rate / Column Flow Rate Split/Flow Ratio 60 cc. / min / 0.5 cc. / min. = 120 / 1 Column Sample Amount that reaches the column. 1 / 121 of what is injected. 123 Appendix E. (cont.) Gas Chromatography Van Dempter Equation - Height Equivalent Theoretical Plates HETP = H + H + H + H p d m s Multipath effect. H . 2 L dp where L is a measure of packing p irregularities . where dp is the average partical diameter. Molecular Diffusion Term. H - 2 G D d gas where G = a correction factor v accounting for the tortuosity of the gas path. where D . diffusivity of the sol§§§ in the gas phase. Resistance to Mass Transfer - Gas Phase. 2 Hm - w dp v where w = a constant of the order 5 of unity. gas Resistance to Mass Transfer - Solid Phase. 2 Hs - qR(l-R) d v where q = a configuration factor depending on the slope of the D phase (film, droplet, etc.) liquid Expected HETP for the 25 Meter Carbowax 20M Flexible Fused Silica Capillary Column ranges up to a maximum of 8170 (Dandeneau, 1979). An actual test run, with calculations based on Methyl Tetradecanoate indicated an HETP = H000 (Anonymous, 1980). 129 Appendix E. (cont.) Gas Chromatography Mass Flow Rate Responding Detector - Flame Ionization. R = K2 (dm/dt) Peak Area Substituting Peak Area Peak Area Integrating Peak Area Rdt where K = a new constant of proportignality. dm = the instantaneous mass of the solute in the detector. where R = responce. K (dm/dt)dt 2 K (dm/dt)dt showing that peak area is proportional to the total mass of the eluted solvent and Peak Area is independent of mobile phase flow rate. Table E1. - Average peak area data f0r all carrot selections included in the aroma study. 125 Appendix E. (cont.) Gas Chromatography Carrot lines and Cultivars (a) (b ) (c) (c) (d) 1A13 1385 1383 5987 107 SpSW SpFN Gde 6000 008. StD. P01 366 17A2 857 1168 151 908 636 676 329 121 667 P02 0 23 O 55 0 0 O 0 0 6 32 P03 0 33 6 A0 0 0 0 0 0 0 27 POA AA8 75A AAl 316 122 318 352 376 17A 609 329 P05 728 1181 8A0 3A6 165 362 957 93A 209 367 683 P06 0 0 0 0 0 0 0 0 2 0 1 P07 8 0 0 36 0 0 0 0 0 0 22 P08 67 107 55 A59 2A 16 139 12A 19 0 113 P09 2AA 512 276 1A23 19 96 503 AAO 0 2A 353 P10 0 0 8A 0 0 0 A1 79 11 0 57 P11 10A 1A3 29 108 0 0 0 0 11 0 75 P12 21 63 59 0 0 0 0 30A 0 0 170 P13 168 A75 20 A362 0 0 280 0 0 0 1632 PlA 0 39 0 0 0 0 0 0 3 O 21 P15 6 0 0 0 0 0 0 0 0 0 3 P16 11 0 0 0 0 0 0 0 16 0 28 P17 0 0 0 0 0 2 0 9 0 A P18 0 0 0 0 0 0 0 0 10 0 5 P19 12 10 0 0 0 0 0 2 67 0 29 P20 0 A 0 10 0 0 22 8 A 0 13 P21 0 6 3 A5 0 0 0 0 0 0 17 P22 5 3 8 29 0 0 0 0 0 0 1A P23 0 O O 0 O 0 0 3 11 0 5 P2A 71 Al 36 35 3 12 0 12 25 26 35 P25 0 0 A 13 0 0 0 0 0 0 5 P26 0 0 0 6 O 0 0 0 0 0 11 (a) (b) (c) Spartansweet Sparatan Fancy Goldpak (d) Standard Deviation 126 Appendix F. Aroma Standards Table F1. -- The ingredients of the standard aromas used in the training/testing of panelists. Aroma . Ingredients Piney ' - Slivers of pine sapling, approximately A mm in width and 8 mm in length. Bark was included on the slivers and pieces. Sweet - Vanillian Crystals - 1 gram Eastman Kodak, Eastman Organic Chemicals, Rochester, NY. Lot-273. Woody - approximately 1 ounce of saw dust and chips from a mixture of woods. Haylike - Fresh mown hay and cured hay ground through Willy Mill No. A mesh screen. Fruity - Mixture of approx. orange extract and ethyl acetate dilluted with water. Perfumy - The unopened bottle of Loren perfume. Ralph Loren Co. Earthy - Various types of moistened soil including river banking, black clay, and sand ground together with decaying leaves, grass, and twigs. Musty — An old book was found which smelled very musty. The book was then compared to the aroma of 25 various mold cultures growing in a collection in the Dept. Food Science & Human Nutrition, MSU. Three molds were choosen which imparted an aroma similar to the book smell: Colvatia M-22 Trichodezina M-ll Fusarium M-20 All were grown on APDA. Scrappings from each plate were combined in a sample cup and well covered with the Dacron Fiberfill II batting. 127 Appendix F. (cont.) Aroma Standards Table F2. -— List of definitions made available to the panelists for both the testing/training panel and the carrot analysis panel. D E F I N I T I O N S CARROTY - Aroma of fresh raw carrots. PINEY - Sharp characteristic aroma of pine. SWEET - A pleasent heavy fragrance. WOODY - Aroma of sawdust or woodchips. HAY-LIKE - Aroma of mowed and dried hay. FRUITY - The light essence of various fruits mixed together PERFUMY - Light, subtle fragrance. EARTHY - Aroma of mixed soils and organic matter. MUSTY - Moldy, stale aroma. OVERALL AROMA - The overall impact of all carrot volatiles. 128 Appendix G. TRS-80 Model II Microcomputer System System Overview. The Radio Shack TRS-80 Model II is a disk-based microcomputer system consisting of two major components: (1) a display console with built in disk drive and (2) a separate keyboard enclosure. The operating system software is loaded from diskette by a built-in "bootstrap" program. The Microprocessor. A Z-80A microprocessor is a the heart of the computer and operates at it's maximum design speed of AMHz (A million machine-cycles per second). A read only memory (ROM) provides power up and reset instructions to the processor. After the Disk Operating System initialization program is loaded from disk, the ROM is electronically switched out of the system and replaced with random access memory (RAM). Random Access Memory (RAM). Memory support for the system is in the form of volatile memory. This RAM comes at a minimum level 32K bytes (lK-102A bits) which can be upgraded to a maximum level of 6AK. 129 Appendix G (cont.) TRS-80 Model II Microcomputer System Video Display. To free the Z-80A processor from display refresh and related tasks, there is included a large scale integrated (LSI) controller chip. The display offers two modes of operation: 80 characters by 2A lines, and A0 characters by 2A lines both of which are used in this application. Displayable characters include the full American Standard Code for Information Interchange (ASCII) upper and lower case as well as 32 graphics characters. The Keyboard. The keyboard also has it's own LSI controller to free the Z-80A processor from keyboard scan and related tasks. The keyboard is in a separate case and is connected to the display console via a built-in cable at the bottom front of the console. Floppy Disk Drive. Included in the Model II is an 8" disk drive unit. Three additional may be added to the system using a Disk. Expansion Unit. A high density recording technique (Double Density) is used in the drive so that each diskette can contain 509,18A bytes of information. It would take a 70 word per minute typist 2A hours typing at maximum speed to fill the information area of an 8" diskette. 130 Appendix H. Personalized Ballot NAME z-” DATE : Honda: - November 17, 1980 OFFICE : 328 Food Science . Please match the characteristic odor from each cup with the list of characteristics below. Take your tine - sou nag resanple as Hang tines as you wish. For your own information your previous scores are recorded below. Thank you. FEFQEEKJJECJLJSS SSCICJFQEEEB TRAINING TESTING PANEL 1 PANEL-2 PANEL-3 I I I I I I I 8/8 I I 6/8 I I 8/8 I I I I I I I c:¢arerac3'r \JCJL_€§1F3EL_EE£5 - 'FF?‘§JEP43CF4C3 ‘rE553'TZEf4C3 CHARACTERISTIC SAMPLE NO. PINEY SHEET HOODY HAY-LIKE FRUITY PERFUHY -_ EARTHY __ HUSTY __ _ ......--.--‘-.8.=BB.B“ .--— Figure H1. -- The personalized ballot produced for each panelist in the testing/training phase of the StUdye H \ o 131 Appendix I. Sensory Evaluation Booth Construction D. Figure 11. —- Schematic of the sensory evaluation booth constructed around the microcomputer system for use in the sample evaluation phase of the study. 132 Appendix J. Data Collection Program XXXXXXXXXXXXXXX MCLELLAN RESERVED AS 1980 R. ETTS .‘ .7 ALLUT/E RII MARK NOV. E ALL 0 0 BY PROGRAM NAME WRITTEN DATE ’xxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxx ’XXXXXXXXXXXXXXIXXXXXXXXXXXXXKXXXXXXXXXXKXXXXXXXXXX ’XX!!!XXXXXXXXXXXXMXXXXXXXXXXXXKXXXXXXXXXXXXXXXXXXX ’XXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXX ’XXXXXMXXXXXX ’XXXXXXXXXXXXXXXXXX!XXXXXXXXXXXXXXXXXXXXXXXXXXXXXXX ’XXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXX I .I I . i I I 0 3 1040 .000 1010 1020 10; 1050 1000 1070 1080 1090 1100 1110 133 30 H h “lib 11“ H hlih llh "TIN H lib HTTH H A N l flllfi H O H h hlifl H m H N “ll" 0 V H N HIT“ fl 0 i H Hit“ N I H H flilflll e TH 0118 h A H H “1!” N C H P fill“ H m H 5 0!!” u v H . “I! u I “Tu HI!Z!l e li_l' HITDIT ’N liH fillkifl c H u I Milk-N a H' . fl {Gil ~r TO N 22“ 9 H J . ITZCKN <2 llO iCJOll e N . liHLLH F~ IIZ . ih- H O h C !l€><fl N H ; llHrfifi ~' ITO I HP-EIH fi 0 P.Z iifiib H i H . n Zw¢ii a 0 o ' :P12 0 H O H H 2 R Q 9 0120 H H H 2' N €121! G N llKmammoeu zmze :ozmzaaa aaaaa asaZHx ea saga mzaz baa; asap azacar.xa.acspzaaaam4o \ 0:0. nunuauauuuu m-xcz «Hmze .mazm _.Hsmzca again. .ii'oi'ee'oiii ‘I‘liti:el.. ‘ x kxmz in..zmm_232 mirqbzcmw £0375: a. “2:. ..n_m._.zm: ......EZH 'p". .aazamziiszzz asizzg 32:. 1:. ;s_:e: .212 ,1... r: _ ... .1.“ '- I. a... .... I... .... .... .8. once 0‘: In. Dee. ': a... on: 000- }: 1|. 3.. lo. '0‘. .... o... ..u. . . .... . .... use. In I. ..l- l..- o-no .01. .000 col a... II. .3. curb Ill. '00. use. so: a... 0.00 I: X: Ive. II.- ‘Iu all. I, .0. '0! no; one. tool a... one. 030 also I... :1. to: .0. 0" To. II- 100. .900 loo .10. :0- .II- cell 0! nul- 00. ovo- clns one. can. In ‘ e u .30. use. a. 000. 1.. It: a... 8.. n I. on I . all! a. I .I to es II 1.! In! etc. cl. - a. gel. .. . at one. .v'l via. 0 o'- OOII II: .... In a u:- a let. II.- as.. '00. In. I: I... T. 00‘ 0.. all. the. no: to I... 'ol see- .nee 'e I... .000 all. '04- ncoe toou Ian. on 'o a... uIOe eel. all. 0': solo a! I 0' ..e nec- I: O0.- 5 a; ZEDHS ”an...“ 3... mm..._.u.£€m at“. Emuwzqaz Zm.I..... ._.3.¢ZH din;Hmiiiai.19 ....uZfin. NIP mmwzwz FELZH .5 .... n. m" . c w. mw .m, cmfid awa— afifiw ccafi O$KH cmmq smmfi =QMm EAH scxa smug own cfikfl coma ooflfl caofi omeq 2 . .... . .... : .... a... F .H. m r, ._. anal c .. ...... Ln scam a a w. A. .....~ m... H a A m H 136 ....H vim u. m.x.S. 2.:3 0011.” A a. mam A. a 1.10 .Eam 01.3. 1:. 30 ._ .1 1.1.. at: 0.2.0: ” . 1.1;. 7.1.... 1 L: . A o . ... H. m. ...ZH ..L .1 L. __1 ._ ....H u A. V 0.51:; 2.: . 1.. : 1H1: Am 00 8 00 000 0000 0000 0'0 00-0 0000 0000 000. 0l0 0000 0000 000. 0000 0000 0.00 0000 0000 0000 0000 0000 0000 0000 0.0 0000 0000 0000 0000 0000 000 0 0000 .000 0000 In. 0 . 0.1 0-‘00II0000000.0000Iei‘eloi‘hii'-00000.0...000-00000000'0'000‘100000000000000000'00‘00'0{£0-0‘0‘000000000000000000-0000!0‘ ”Mo. « NNNNNNNNNNNNNNNNNNN ZCHFWMDG >PH$ZMPZH NNNNN~ GKQN NNN!LNPPNNNNNNNNTNNNNNNLNNNNNNNNNNNNNNNNNNNNN. GQDM . amcm anfiIuA: CL. A ion... «21:... Am. .A: ..L...LIU FZHNLL»AomvaEIuLZHiLumJu -mwcm xxx mozoumm 3mm xxx Empty; .NLZHP . ”x ...xwzncook Auhfinx ..LOL 9...“...7... AANNvmnsvmm A\A. .~va.L.ZH.vL.u—. . ...H amatzz mJLZQm...Am.\.vm:.ZHHLL 3:7. ..hxmz..L.Z.n..LL mm-.m ...—..wMHL: .PZHJLL 2.5; MANN LHuu: EDD>zaAm1mvm...—ZH....L”:.:.AA: ...ANXV$.ILLv~Lh.wZH 1ANxvan—L;LLM_ ..:. Am.m.vm:.ZHE..: A- .ALV¢ZIU._.ZH.’L...L AnmyfiLIr:Z~J1_ of...” cm OHAHNN ..LOL amt“; ANNVZmNANNVI. X40AmALZ 31.3.; \ 3N3 .... NH... utflzhxhnnfi..n.u..ri.a.ufln... 063.! anaaaar on_.o¢:1w_zH _.Hauz a: m aH:Hz uaaaar. ass. 000 0000 00.0 -00 0000 0000 vol 000. 000. 0'0 .000 0000 0000 0000 i‘lt‘Iil‘I" D'I.'U‘|.!!!"U N 0001 0000 000 0‘0 ..0 0 0 0 0000 0'00... 00 000 | 000000000000 0'0 0000 .000 0'0 "000 0|: 000“ ”"0 0000 i'o 00000.0..- 00 0000 1000-"000“ “own” “I 000“ 0'0‘000'0 N N 0000000 :0.“ N 0000 In. N NN NNNNNNNNNN. DTfifl . onfie AIS: _..—c.113. x .._.. .wz mm..§ aaaa 222A o 1aE1 1.:1a_1.1_azH 1H. ma opan. a2. ani1 ,. 1:... 0 0 . . 000 0000 0.0 0000 0.00 000 0 000 0000 -00 0000 0000 0000 0000 00-. 000. .100 0000 00 00.. 0. 00 000 000. 0000 ...... ' 0000 0000 .0'. 000 0.0. 000. 00.0 0.0. ..0. .d .d . - . ... 1. v 0000 .000 0000 0.0 I. 0‘0 0.0 0.0 0'" 0000 0.. 0000 0000 .000 0000 000. 0000 0.0 “.00.. 000. 000. 0000 0000 00 0'. “H00 00 00 0000 I00 000 .000 0.0 0000 000 0'” ouun ."0 0000 0000 0000 000. .00 nuaauaa chz I _1.:c1 aoa gauze aaaar. ;;1 . .... .... .... .... .... .. .1. .... .... .... .... ..0. .... .... .... 0.. .. ...a .0: .00. 000: .000 0000 ...0 0000 .000 0000 0000 0000 :00 no“ “0 .8“- .n.00 .000 00 “000” .0” .00.. 000. ".00 000. .u.. 000.0 .000" 000” "“0 .000 .00. 0000.0 0. I 0000 0000 0000 .000 0.00 0.00 0000 .00. .3. .10 ..00 .... . ”.4 .0 a :37. c :: 2&2» Aaamr ax aaaa+axr.x. xxa1ex1mm can1 0000 0000 .I0 0000 00-0 0000 0000 0! 00.0 0000 0000 000. 0000 0000 .0- 0| 137 000 00 0000 '0 I000 000. 00. 000 .0 0000 I. .000 0000 00.0 0000 00.. 00.0 000. 000. 0... 0.00 0.0. 0000 0.0. .000 0000 000. ..00 .000 0-0 ..00 00 .0: 0 00 0 0000 000. 0.00 000.0 0.0 ’00 0000 “00“ 0h” 0.000 "00" 000.0 00'. -0. '00 '00 0000 '00 '0 000. 0000 0.0 0000 .00 0000 0000 0000 0000 0000 i 0.00 0000 0000 0000 0.00 0000 000. 0000 unnuuuuununuununnunnun“ ZOH—Hmdna ..PLEH _—A€¢I~ Wm: ......x....2.x I 00 '00 0.00.00.00.00. 00000000.; .00. .000 0000 -000000000 000..000000.00 .00 0000 0.0 0000 '0 '00 0000 0-0 0! "”0100 00 00 0-0 0000 0000 .00. .000 0000 0000 00.0 00000 .000 0000 0000 0000 .0000 0000 0.0 0000 0.00 0000 00 000 0.0 0000 0000 cxum DbDfi aim...“ 3.5.5 I mN.§AxnuAA...IVI.AIA..-AXHAX. ..x...AA.L 30:237. 2.. ....IIF .. ......th LH chum... 01—30. Aam A v4. 1: .I. A AX . AI v EPZHLL .MIIH...I.HIAI~...A......... ZwIIL. : . ...“..mwéx .LH AIANA: zmIL. AIAMAX ..IH.A.:AXNAXI I: ...AAx.mIV....I.I.I.IH...I.I ZHIIL. .....NfiVI .LH ck-.N 3.86. AOWA v ALI—ID. A A.x . m V AILIHVLL .mImeImIII: ..-.“ ZszL. .. . ...mwéwI LH II... ...... ..-Ax ZMIIL. omhAx LHH A+.A.. “Ax” I .. .. . A Ax ......LmILInguIahI ZMIIL. .. . :utIGI .HIH AI Kama 244:. ....NGVI .IH»9>NIVIZHN$VI on...“ Ax. Aam AvalII AamHJMIVmCIHWIL \ x 1-....- NPNNNNNNNNNNNiNHh.ruh.NN.vNNn.NH.uhuv"...“..hvhxu \ NNNnunuN .. NIP WUflIi—fi— .... _..mtu :2 3°“. C .vfiad.v.nus AmquA v fivLIU. A on». kaIPZHNIL X ._.sz“ A A cmA v 4.210. AX . AvaPZH—“KLIQILUPAGNX ESL ”AIsmHA v ALHIIU . A emu. RVQPZHEL x .FmeZ” A AAIWA Xvi—IO. Ax.kaI.—.Z.A.NIL.$N 3L. AANX EOL ammAvfimIIU. AAIA \va ..FZHxIL ..>.FHmwz.m_.—.ZH... Am..ugmvmv.—.ZH.&.H. ..IIJHII: . AKA—0. vaIL.~I.I.zI..u: . :ZDHAHwZ: . ARM . .wva._.ZHLL» “:13. -..: . AAIA. mJQFZH ..LL AHmN.I....IIU.: :.A-. .xva...L.uI.: ...Ac...vI..I.LI..IA:I ....”I Z Z : maswLIQ... : .AzANNVm.m..vAw ..L... :.A.VN «m.x—IS...” m. .I Z r... m.. :AADNVA....IIH.I u: .. s vanwio .. .. » Amwhsvfivvmztvo 2» .n. o 7.— d: A A0 Amvm~.—..Z~..u&.& 0 Ln...“ Awnvmwam. $30“. mmAwLQ ...—Kw ..ALAwt .— . .. . A o . A VELZHNIL .. ... 2 AI G In N a N. “..N I... A ...... A... I. AI ...m ....H a a... ....u ..H... AI ...w m... m... A...\ «... ... I... 3 ...... ...... I .. ....“ .... ...; cemm AI Aw N N AI «u m. w.” AI A. ...... .... comm o A... A N a a A ...... cmAN o C A N o ...... A m... AI A. A M... A... A.” A A... A”. N. .I.. _..... 138 .A: 13...: ...... ... mmwmwga . . . ..AA. aHvaz..- ...?Q... .. .A... quezT—u“: .w .. .39 ......r.....:;...m ......u...“ 00-1.5. 0H; .Ac.qmveyzam¢ 0. ‘~u II.- ..t a. on.- . . D a»). O‘- n ..I O. ..I. O .. ~ ... .u -.. A. ....H .... ...: Ti... .... ... ...A A... .... .. ..“A A ..... . ... ... A... .....h Aw. A. A“. ....H... 3 3.3.“... D A .33 "mm.” . .kH ...... 3 .. .... .... _..A. .... H... .H. >3 HI.” ...x... .. A .H...» .. V H A m, ..A 2.74.”. ZMAIA. .. .13.“. ....A ...H .... NMN D A. 8.4.. . A mm...H. . Him—AU. A.HX. mHV m. .PZH ..H... W-.. .mAAuAWm ZmAIP .. . ...Anwéx ...AH A.muu.—HX z ...w—z—F OHM”: fix ...—3p...» His”. ”3...: .uu .. .. nwAXA~mfivme—2Hvu.u_ ZUIP .........n....$w.- ...—H chN C .A. Om. Amva V wZIU. A HX BH v EPZHKR mmeommN ZMAIP .. . ... ”we; .uAH 2.“...on ZUH: an... Hx ..AH. H+Hx:quA. : .. . A Hx ......H vaHZHEnA ZUI... .. . .....flfix ...AH cmfiHx. . Am.mH Vanni“... . A on . mH ”my—.znmzi S . 0‘. '0 O... I i C 'I I ' 3‘ C. 3. '0 V... '8 O' 0" I O 'I -0 I‘. i. "A i '. ... 9! ‘ I! t ’0 .... .-. "O 0" cl. - '0. '0 ‘9. Q... g " -.. ... k i ‘ I i O. t i '5. ‘I .9. 0‘ i I'D ‘5 i i l I. ’0 l. i I! .0- i " 9... " '0‘. -0 II out: II' ‘ 0. IO, “ '0 0 'ID OI: - t. I. ‘0 C ! ... 0. 1‘0. -.. 'I- I.‘ ‘ OI. Q... 1 !. ' " i '0 O . . I. O ‘ OI I n I : l I ‘ s O . u I. o n .0. Il.‘ I... I'D ' '0. II I. 0 ll ' - I... ll 1 '- 0'. ..- '0 .1, ‘ll '0 i g ‘0 .D '0. .00. " ' O... , .... ‘9. ' ‘I. ' 0* I... '0.- ‘l 0 IO. I... O... 'I C! i ! l. i ‘ I C. I ‘0. .0 i " I: II 0.. C! t I ‘ i I! O! '. .' 0" I. O! 0‘ ... .00. '2 ' a I: c '0 u ' II ‘ x bxwz: x .mez” \ A..H:m.Hz a m .-. A. : . Acm . AwH V M.A.ZHZnAA AmmH vfiazu . A cm... . A~H VapZHzL A AABH :3..on . Ax . A. H ghzumai A .wAuAU.A.Hm3x ””5; A A va v 4.9.1:... . A on . cH ..m..A.ZHH..A.n.A A A cmH Var/.25 . Ax. . vH yahzfiuiuau OPH Hfix as; “AmmH vwmzu. Ac H . ¢H VmAZHzi ..>.A.H-AH"w¢.mHmma: . Aw mH yahznnzi zacogam COP: . A HA. .mH v.3. :— 2H2..." ”.....me 08.7.. 2:. mH $.21: . .1 .. J. _.d Jun- ’V ”L :....:. ”Mafia...” c {mu/... H. hm“... 9 A1“... A. M. ...u ...u cum...“ O H ..w. b. comm : ...» A... ... a ...w H. ...... ch¢N O£$N o A...“ H. M... A. H. Au ...... c m. r. ..H... _... ..... ...: .... c H v... ... o a e. ....H A”... a... ”u «.... ....-.on «:3...» 5.39. . . . . otAH OHOQAXmeoncsoH Oanx men» m>m¢H an: new 30: misc mgm 30> .Domlscmz».: unANXvwviiu: .Am.NHVmPZH£& mammmmzz.Am.ch@b2HzmumJuuNN hxmz AcNVIAAaHVIAAmHvIAAmHVIA AQHVIAAWHVIAAchIAAmHVIAANHVIAAHHVIAAaHVIAAaVIAAvaA ARVIAAGVIAABVIAA¢VIAAmVIAANVIAAHVIAAmNVIAANNVIAAHNVIA i!!!’i!i!tli‘!: 139 nuuun mgmzqm A.. munwfiuaAAw mez :AAchwzzuA «toad hxmz an. oceu :# u! I... o"..- .l 0:. vi... ’0‘Io‘n‘i‘o!‘ ZMIP mwmm. ._ a. #.#*#:QZHm3.N#PZHai :DO>IxZ€Ih:.Amm.NHvekZHiaux mez AA:X:.A>ANV$PZHKLAAQKVQZKH>AANNVszuNuccH Oh Hux mounmgu quAvaIAAANNvmmvzmuANszuciuAHNvI . N hxmz seem Show cHnN zmzh .. :3 VA .2. ocHN chow zth :\:fl¢x uH ....nufix LHu$>UXZHH$VA : .LmnA :«AvanbAHAU ... C0 @DCM9500 OF: chN scam caku owkm onkm comm J I OI” mun- N. K3 ovum omnN oNAN :HAH aokm omom emflm oucN swam 1130 u... .90. I}. It. s !I tall o... 'I- .-. o! .10. n... o‘- 0‘: c.- lo .0 I '0- : v0 ‘0. u ~00 o! cal I... an. ...I no: I! .... cic- . .00 on: :3 vv-. ...o u... I I. u:- on.- a... 2‘. ulc- I I .‘o :l- .. 2. an: 0:- .0- II. » In. u... noon to. -\o In. ul. ul- .... II- I: out. .0». II o V's «I. ...-o II». ..o lo. 0:. cl. .0- z. i I I... ‘00. cul- '. ‘0. nova to!» .000 Out. .0. o I... I! u-- I... .11 '- Aw VI- ‘I. «:00 cu. .A.Z voo- .l- 1- ul A”. 9!: not. :1. I: Z I... I... w cl. '0. A. v! .... .... O .H o:- .l- ‘0 II. .A. oul a! .w A .A 2!. .lo 3 vol- 3'- Z I‘l- .5... c a! to: Q... '- Ky. 03. o. a. .10. H o..- ton. cal .00. C ..X 30‘ ... [I .0. nun. . v: m ...-I 0.] 2.: p:- X.- o... ...— .. ... In: 3.. a-.. 3.. .... In. v00. .1. ul- vol .-. 0:... ...- A... no.» up. .II cl: on: I". z I... I. u! :I In. ava- .I: ....A c... not. al- snow .10. .-. A. a... not. .-I I! ....A n I. '0'. '9‘. u'. :Q. 00-. o... '0. . o... on! no: -.. :3. on! .... to: .m— ..X 0'. O... t.- A. x.“ .... ...... .... .. .... C .V .... ... c Aw A..." ..u c A. a ...... 141 Appendix K. Mass Spectra of Some Compounds Found in the Headspace of Raw Carrots 142 ... z .. mm @3056 so.“ “manages 0:... :35 .3 Java up 6:50me mam: II .HM 93»:— {E 8. 8. o: 8. 8. c: on. 2. a: 8. 8 8 2 8 3 c. an 2 . _::_::_::_::_..:_::_: lc __ __ _ 2 s... ; Tanya m . T9 w z... 955...; 13m 9 N new =o\ //xu\ :9 Io“ _ é N L... 5 ....... .. r8. \ “32.23:: 8.22:: 300 2:32; 293‘ K/ 5 M5 143 02 8— c: .: z : ma wouoamv cow umaaooaoa any saw: N‘ xwwm mo asuuuunm mam: II .NM auuwwm {E 8. O». o: On— ofi o: 8. on 8 2 3 On 9 an em Azttcfic. «>223: 300 2:3an :32 144 8. 8. c: 8. 8. o: 8. 2. Egg. wo__:__:_wo__: :. ... z : mm wmuocow so“ Haasuwaoa .23 :33 .1. xwma mo Eauumam mam: ...I .3— wuswflm {E O: 8— On 0. On ON _...HEE la lo. r2 Tonya 19W 118m 1.8m T2 a VB loo a.“ 25:5 .. Q ~ ~4\u/:u\u z N 5 ...... -5 [8. 0\ “32.23:: rtotzv 300 3533:. :02 M.au/\ :5 145 ...z: 3 @383 Goa uwfisowaoa sags m¥ xmma we aauuumnm mam: II .eM ouawfim oo— 8. o: 8. On— 01 On— ow— 9.9.3.3 a: {E 8— 0o 8 Os 8 On 0. On Ow :EE: 1 .... Io. Ten Tecw Igwm Tosx Tea ina r9: A.£:.c££ :23qu 300 .232; 203‘ 146 .: z : mm wouocov :oa umasuuaoa mzu :ufi3 ¢* xwma mo abuuuwnm mum: II .mM muawam {E “a. .a. o: .5. 92 o: .3. o: o: .8. fix rbrttrruuprtpEEttbtptzraEztczrzzcgrszatbrtnzzJ 8 9.3.22 Nate: :9 ./wnnlu&“ x/onuo:. \ ~51: m as Os 8 On :_::: __:::: _::::__::_:: o. On o~ 3% _ 0 Yo. .....ow TO. N ram Iquw TON r18 T8 Fae 73:23.5 untottv 306 2352*“ :0: 147 .: 2.: mm umuocov :oa “masuwaos wcu suas Haw xwwa mo aauuownm mam: II .mx muswwm {E 8— 02 O: 8_ 8 8 as __b____._.___.__._.__...___._._ oo- 8. a: 8. On. 8 On 0. On 2 __._bb_b.____._.__...—....__..__.____.__.__.._ o: 2‘ 98:05: N ~ m 5.1.19 x 5V T5\ V”. in: ~=o\ Wulux — ——= q o Io. T8 ..lOnub T. ... mw 18m ISW Tosu, To» 1.8 r8. «2.22:3:— o»..£ozv 300 .332: :02 211553 oo- 02 o: 8. On. 0: On— _.____...__ _____ ..____» pm___._...__.___________»__b____._..____ 9.9.35.7 » m ..n.........ny~._ _.Hvllll, \A“v III“. m 5|! u/ \\ lg .. :a\ ~51: _7 ow— .: z : mm vmuocov cow umaaomaoe ecu nu“: mH* xmwa mo aauuuwam mum: II .sx ounmam {E o: 8. _E::_::_::_ _ 8 8 _.._.________ 2 o. 0n 2 _._.____._._.._.____._.._____..—_.________.__.__.______. 1% _ _ ;_ __ _ .. T. o. 9. an TOvW 8 Tag” 18“ T. ..n u. 78 Too r18— ?oEEoE. 332:: 390 3392“» :02 149 Appendix L. Calculations for The Coeficient of Concordance, "W". Sample Totals : 48 29 37 4O 41 Grand Total = 195 2 ( T ) = 2304 841 1369 1600 1681 7795 2 Correction Term (C) =(1951 / 13 x 5 = 585 Samples 7795/13 -585 = 14.61 2 2 2 2 2 Banks 13(1 + 2 + 3 + 4 + 5 )- 585 = 130 Samples - l/n w = —— ---------------- = 0.119 Ranks + 2/n ’ W(n-1) O.119( 13-1 ) F = ———————————— g —————— ' 105119 Degrees of Freedom in the Numerator 3 (5-1) - 2/3 8 30814 Degrees of Freedom in the Denominator - (13-1)((5-1)-2/13)= 46.08 The F VALUE at 5% for 4 and 46 = 2.57 The F Value indicates insignificance. The Judges were not able to agree in their ranking. — —— _— Same analysis was completed for the second set of five samples with a calculated F value - 2.29 again falling short of the F value for the 95% significance level. 150 Appendix M. Motivational Liturature Michigan State Univ. SniFFer’s Associaticn N E W S L E T T E R Editor : Mark McLellan Date : Nov.2?.l?30 Letter Frgm m figjggr: Early in 13980. plans were laid For a special “taste panel“. This taste panel was to be the mainstay oF an analysis oF the volatile constituents oF RAW CARROTS. The ground work For the panel was not introduced until mid 1980 when it was decided that the panel was to be trained and tested in it‘s ability to distinguish between smells similar to (but not necessarily identical to) those Found in RAW CARROTS. On October 14th. two Open Discussion Panels were convened to discuss. gather. and develop descriptors For ten largely diFFerent varieties of RAW CARROTS. AFter renewing the comments and recommendations oF the panels. the Following ten descriptors were chosen For their Frequent use and conmon agreement between discussion panelists: Piney Sweet Woody Hay-like Fruity PerFumey Earthy Musty On November 12th some thirty seven people were asked to sniFF ten standards related to those Found lay the diswssion panel to be important. Each panelist had a COMPLETED ballot beFore him/her and was told to identiFy the volatiles using the completed ballot such that on later occasions the panelist may be able to complete the ballot him/herself. This was the only time all of the volatiles were identified For panelists. One day later all Forty panelists were tested in their ability to separate and identiFy the ten aromas. Five oF the Forty panelists were dropped From the panel due to poor scores (more that one wrong). OF the Five dropped. three were older then the average age oF the panelists.another was a Full time smoker. and the last was oF the average age oF the panelists. ' One day after the First test. the second test was conducted. with thirty Five panelest remaining.Seven panalists were dropped From the study due to their scores (less then perFect) and three others were dropped due to availability. Three days aFter this second test. the third and Final test was run on the remaining twenty two panelists. OF those. two were dropped due to their scores (less than perFectl and two were dropped due to availability. Eighteen panelists remained and constitute the trained and tested panel. The panel consists o‘ 567. Females and 447. males. QEEQEKEEHDN Data collection is now be perFormed on the TBS-2'0 HOD II microcomputer. All measurements are made by the computer. coded and stored For transFer to the Control Data Computer on campus. This is a great help in reducing time For statistical analysis. The average time to take measurements and code the data on one ballot is 5 minutes. So assummg at any one panel only three samples are given that results in 15 minutes oF measurements per panelist . throwing in a two minute rest between panelist pruduces a total time or‘ 17 minutes per panelist times 13 panel members. The result is put over 5 hours data preperation time. This does not include the additional time For keypunching and data checking which could easily run the total up to 8 haurs total. All oF this is Figure M1. -- A Newsletter distributed to the panelists as a motivation for achievement. 151 Appendix M. (cont.) Motivational Liturature accomplished instantaneously by the microcomputer when each sample is completed by the panelist. requirering no additional manipulation. recoding or rewriting oF data. mmsa All panelists are reminded that when three samples are given to them. they are to treat each individually. DO NOT compare between samples presented. waétgggys Starting with the next panel. the computer will make an extra eFFort to remind each panelist which sample is to he sniFFed and also let you know in no uncertain terms when you are done with the set oF samples. emaomm I’d like to thank all oF those who are on the panel For your considerate donation 017 time in this study. Your help has been and will be most appreciated. 'How long will the study run 1’". you say. I hope to have all sniFF panels completed by the second to third week in January. Figure M1. -- A Newsletter distributed to the panelists as a motivation for achievement.(cont.) 152 Appendix N. Factor Analysis Results Table N1. —- Correlation coefficients for the factor 1 analysis of sensory evaluation data. Sensory Parameter Q01 Q02 Q03 QOH Q05 Q01 Q02_ .821u0 Q03 .01522 -.Olhhh QOA .0h218 .07h86 .50299 Q05 .38199 .2677“ .01992 01669 Q06 .u12h8 .h1896 -.18700 -.03908 .51h77 Q07 -.08568 -.10173 .uzoss .2988u -.08188 Q08 -.O352S -.03932 .19093 .3h505 -.O906l Q09 -.09087 —.O7730 .39030 .22u06 -.OO679 Q10 -.08795 -.06582 .19h63 .3521h —.08979 Q11 .15N53 .08875 .281h1 .1228h . .377h0 Q12 .30615 .3hu30 -.22fl70 .0590“ .l7h85 Q13 .00395 -.09173 .31873 .0922h .2fl268 Qlfl -.073OH -.oauuo -.18355 - 03135 .1N086 Q15 -.llh70 -.11932 .366H5 .18169 .03099 Q16 -.00688 .0263h .15597 .2h9h0 -.0862H Q17 -.O9766 -.12523 .38122 .20966 .lh132 Q18 -.Oh70u -.O6l66 .13616 .21817 -.00232 Q19 .6l7lh .h994h .12179 .20722 .322u5 Q20 .h2532 .h5239 .09816 .2h215 .2023h Q06 00? Q08 Q09 Q10 Q01. Q02 003 QOA Q05 Q06 Q07 _.32uu1 Q08 -.19582 .531u2 Q09 -.28953 .7lh67 .392h8 Q10 -.1935h .h3532 .68802 .63u98 Q11 -.Ol90h .16015 .00506 .23216 .03991 Q12 .50745 -.30535 -.11833 -.31350 -.06985 Q13 -.lh800 .28070 .0750? .3H562 .08053 Qlu .28763 -.l6567 .0h673 -.lh76h .05216 Q15 -.338H5 .60319 .21722 .6h279 .28590 Q16 -.lll69 .h002fl .39870 .37808 .h7795 Q17 -.27362 .526hl .2h893 .62h15 .2902“ Q18 -.07h58 .26975 .NH668 .32229 .h6ll8 Q19 .3fl129 .03617 .21953 .0665? .2322u Q20 .27505 .15363 .29832 .19213 .35396 153 Appendix N.(cont.) Factor Analysis Results Table N1. -- Correlation coefficients for the factor analysis of sensory evaluation data.(cont.) Sensory Parameter Q11 Q12 Q13 th QlS Qli Q12 .12036 Q13 .51390 -.1h878 Q14 .08069 .39h58 .30936 015 .27812 -.38341 .31568 -.2615u Q16 -.02956 -.08659 -.01358 .00fl73 .53771 Q17 .3h235 —.35869 .h5272 -.109u9 .77166 Q19 .12075 .2h951 .06fl25 .05826 -.02513 Q20 .11196 .22202 -.08fl01 -.0h580 .0520? Q16 Q17 Q18 Q19 Q20 Q11 Q12 Q13 Qlu 015 016 017 .30633 Q18 .A9753 .h8357 Q19 .lflh28 .06959 .23670 Q20 .213h2 .10158 .31536 .7h368 154 Appendix M. (cont.) Factor Analysis Results Table N2. -- The Factor Matrix Using Alpha Factor for ‘ Sensory Evaluation Data. Fact(1) Fact(2) Fact(3) Fact(h) Fact(5) Q01 -.12133 .76h6h .07H78 —.3fl613 .09633 Q02 -.1325u .70396 -.o3007 —.33776 .06169 Q03 .52973 .0h90? .25662 -.20823 -.u3806 Q0“ .427h1 .21596 -.083?9 -.20823 -.N3806 Q05 -.0h677 .50922 .fl113h .089h9 .13581 Q06 -.388u0 .61305 .01fl91 .13935 .08H91 Q07 .755?“ -.06892 .02520 -.10516 —.03987 Q08 .60789 .11389 -.362u6 .13911 -.11920 Q09 .78603 -.02826 .09868 -.0556h .08178 Q10 .67731 .13321 —.38002 .18695 -.08005 Q11 .2h306 .2517h .57162 .09558 -.08939 Q12 -.35u85 .h927h -.08253 .23682 -.1h982 Q13 .36538 .0fl826 .62899 .266h0 -.06379 th -.12180 .21218 .0360? .8537H -.0h032 Q16 .5518u .090h3 -.29363 .05991 .17512 Q17 .7h113 -.03299 .3556? .007h0 .27191 Q18 .5h827 .1h962 -.2h352 .28390 .21382 Q19 .16302 .76323 —.11953 -.09710 .02702 Q20 .26h29 .67250 -.25h83 -.117M6 .03803 Table N3. -— Explained variation for new sensory data and peak area used in the second factor analysis. Factor Eigenvalue Pct of Var Cum Pct Cum Pct (based on) (based on) (based on) (Total) (Total) (9 Factors) 1 6.6h602 21 21.h 25.3 2 n.98273 16.1 37.5 #2.? 3 H.15600 13.fl 50.9 59.1 n 2.93359 9 60.u 71.1 5 2.08736 6 67.1 79.5 6 1.8917u 6 73.2 87.0 7 1.h0087 h 77.7 92.2 8 1.16378 3 81.5 96.3 9 1.10497 3 85.1 100.0 '155 Appendix N. (cont.) Factor Analysis Results Table Nu. -- Varimax Rotated Factor Matrix for new sensory data (previous factor variables) and peak data. Fact(1) Fact(2) Fact(3) Fact(h) Fact(5) VBl .09831 -.OOOOH -.08039 .8311" —.00985 VB2 -.08172 .18386 .0017? .01u18 .20398 VB3 .0953? -.0fl371 -.136hl .8022” -.05h95 VBD -.1197M .08285 .12211 -.63&61 .06513 VB5 -.l6916 .01798 -.13569 .52658 .20821 P01 .0237? .4382? .1h963 -.18l30 .08595 P02 -.03972 .78380 .03712 -.08320 .30737 P03 -.03269 .66078 .1331fl -.11h75 .53908 PO“ -.10538 .05358 .26515 .10191 .38090 P05 -.095u3 -.02590 .82888 -.01277 .37963 P06 .96319 -.01919 -.04839 .0566“ .0107? PO? —.Ou083 .05110 .9698” -.O6??5 .1235? P08 -.0523h .32286 .90515 -.08011 .16799 P09 -.06308 .92107 -.0699? .03207 .ozuuu P10 .05792 -.06693 .73277 -.06161 -.l8953 P11 -.02006 .h222h .1510" .06969 .80327 P12 .01331 -.05630 .67370 -.13012 .03337 P13 -.0h969 .910h1 .Ohlh? .02902 .05333 Plh .03h36 -.00662 .15661 -.08030 .92235 P15 .0190h .0180? -.01h60 .h0092 .12658 P16 .80931 .00799 -.0h306 .2969u .12216 P17 .90615 -.0575h -.12651 -.10682 -.05102 P18 .96319 -.01919 -.0u839 .05664 .01077 P19 .96186 -.02737 -.05390 .07908 .01179 P20 .12006 .19572 -.11179 -.0u028 .07050 P21 -.0HD65 .33295 .01897 -.09H96 .lh29? P22 -.ouruo .81996 .03231 -.05375 .07318 P26 -.03028 —.13301 -.Oh676 -.OH75H _.0102h 156 Table Nu. —- Varimax Rotated Factor Matrix for new sensory data (previous factor variables) and peak data. (cont.) Fact(6) Fact(7) Fact(8) Fact(9) VBl -.05625 -.128N5 -.06238 .03025 VB2 .oouu9 .685H1 -.01976 -.l2826 VB3 -.033?8 .17280 .08906 -.05802 VBN .0u016 .3fl286 .110N3 .06686 VB5 .03305 -.095h6 -.15380 .38755 P01 .00679 .19033 .3113? .1661" P02 .18065 .00212 .28986 .257u2 P03 .15hu6 .02729 .28190 .213?h PO“ -.00913 .09192 .62326 -.13532 P05 .018u9 -.0fll21 .32299 -.06703 P06 .00071 -.01?u5 .0H86? -.06581 PO? -.0328h .0658? _.ou991 .Olfllh P08 -.01?H9 .11h30 -.10836 -.01007 P09 .0fl276 .09010 -.20223 -.09016 P10 -.OHO38 -.00886 .23266 -.133?5 Pll .05032 -.06289 .066h9 .0955? P12 -.01855 -.05543 -.00146 -.00300 P13 .01338 .16653 -.25uo3 -.0857l P1“ —.00??0 .13001 .09016 .01520 P15 -.038?8 -.58191 -.1u777 .0522? P16 -.03085 -.fl8h27 -.09556 -.001?2 P17 -.080h8 .20280 .02165 .28709 P18 .00071 -.017u5 .Ohh6? -.06581 P19 -.0h598 -.10171 -.0h487 .0136M P20 -.01323 .1352“ .00552 .510“? P21 .93212 .0260“ .07238 .ougzu P23 -.00?05 -.00303 -.09352 -.01629 P2A .07??? .06831 .063H9 -.26020 P25 .78915 a.027u1 .09737 .07386 P26 .88666 .03869 —.1097h -.11086 LIST OF REFERENCES Alabran, D.M. and Mabrouk, A.F. 1973. Carrot Flavor. Sugars and free nitrogenous compounds in fresh carrots. J. Agr. Food Chem. 21(2):205. Alabran,D.M., Moskswitz,H.R. and Mabrouk, A.F. 1975. Carrot-Root Oil Components and Their Dimensional Characterization of Aroma. J. Agr.-Food Chem. 23(2):229. Amerine M.A., Pangborn, R.M., and Roessler, E.B. 1965. "Principles of Sensory Evaluation of Food". Acad. Press, New York. Anderson, C. 1980. Personal Communication. Michigan State University, E.Lansing, MI. Anonymous. 1970. Vegetable Production Recommendations. Ontario Ministry of Agric. and Food, Publ. 363:72. Anonymous. 1980. Flysheet accompanying deliver of Fused Silica Capillary Column. Hewlett Packard Co. Boyko, A.L., Morgan, N.E., and Libbey, L.M. 1978. Porous polymer trappings for GC/MS analysis of vegetable flavors. In "Analysis of Foods and Beverages Headspace Techniques" p.57. Academic Press., New York, NY. Buttery, R.C., Black, D.R., Haddon, W.F., Ling, L.C., and Teranishi, R. 1979. Identification of Additional Volatile Constituents of Carrot Roots. J. Agr. Food Chem. 27(1):1. . Buttery, R.C., Seifert, R.M., Guadagni, D.R., Black, D.R., and Ling, L.C. 1968. Characterization of Some Volatile Constituents of Carrots. J. Agr. Food Chem. 16(6):1009. Carlton, B.C., Peterson, C.E., and Tolbert, N.E. 1961. The effects of ethylene and oxygen on production of a bitter compound by carrot roots. Plant Phys. 36:550. Carlton, B.C. and Peterson,C.E. 1963. Breeding carrots for Sugar and dry matter content. Proc. Amer. Soc. Hort. 2:332. 159 Carolus, R.L., and Ells, J.E. 195?. Bitterness in Carrots. Incompatibility of apples and carrots in storage led to discovery of cause of bitter flavor in carrots. Amer. Veg. Grower Feb. Chalutz, E., DeVay, J.E., and Maxie, R.C. 1969. Ethylene- induced isocoumarin formation in carrot root tissue. Plant Physol. Hfl:235. Clark, R.C., and Cronin, D.A. 1975. A new technique for trapping and sensory evaluation of flavor volatiles. J.Sci. Food Agric. 26:1009. Condon, P., and Draudt, H.N. 1963. Production of 3- Methyl-6-methoxy-8-hydroxy-3,h-Dihydroisocoumarin by carrot tissue. Phytopathology 53:1119. Cronin, D.A. and Stanton, P. 1976. 2-Methoxy-3-sec-butyl- pyrazine - an Important Contributer to Carrot Aroma. J. Sci. Food Agr. 27:105. Dandeneau, R., Bente, P., Rooney, T., and Hiskes, R. 1979. Flexible fused silica columns: An advance in high — resolution gas chromatography. Am. Laboratory. 11(9):61. Dawson, E.H., and Harris,B.L. 1951. Sensory methods for measuring differences in food quality. U.S. Dept. Agr., Agr. Infor. Bull. 3A,1.- , Dodson, A., Fukui, H.N., Ball, C.D., Carolus, R.L. and Sell, H.M. 1956. Occurrence of a Bitter Principle in Carrots. Science 12h:98h. Ezekiel, M., and Fox. K.A. 1959. "Methods of Correlation and Regression Analysis", 3rd Ed., Wiley and Sons, New York, N.Y. Gipps, P.G. and Casimir, D.J. 1973. Scoring of taste test data on computer cards. Food Preserv. Quart. 33:15. Heatherbell, D.A., Wrolstad, R.E., and Libby, L.M. 1971. Carrot Volatiles. 1. Characterization and Effects of Canning and Freeze Drying. J. Food Sci. 36: 219. Heatherbell, D.A. and Wrolstad, R.E. 1971a. Carrot Volatiles. 2. Influence of Variety, Maturity and Storage. J. Food Sci. 36:225. Heatherbell, D.A., and Wrolstad, R.E. 1971b. The Enzymatic Regrenation of Volatile Flavor Components in Carrots. J. Agr. Food Chem l9(2):281. 160 ’ Henderson, D. and Vaisey, M. 1970. Some personality traits related to performance in a repeated sensory task. J. Food Sci. 35:“07. Kendalé M.G. Rank Correlation Methods. Griffin, London, 19” . Kramer, A. A rapid method for determining significance of differences from rank sums. Food Tech. 1h:576, 1960; 17(12):12u, 1963. Lester, G.E. 1980. Physiological and quantitative determination of differential sugar accumulation in carrot (Daucus Carota L.). Ph.D. thesis, Michigan State University, E. Lansing MI. Lester, G.E. 1981. Personal communication. Michigan State University Liken S.T. and Nickerson, G.B. 196A. Detection of certain hop oil constituents in brewing products. A.S.B.C. Proceedings ,5. Linko, R.R., Kallio, H., Pyysalo, T. and Rainio, K. 1978. Volatile Monocarbonyl Compounds of Carrot Roots at Various Stages of Maturity. Z. Lebensm. Unters.— Forsch. 166:208. Martens, M., Fjeldsenden, B., and Russwurm, H.,Jr. 1979. Evaluation of sensory and chemical quality criteria of carrots and swedes. Acta Horticulturae 93:21. Murray, R.E. and Whitfield, F.B. 1975. The Occurrence of 3-A1kyl-2-methoxypyrazines in Raw Vegetables. J.Sci. Food Agr. 26: 973. Otsuka, H. and Take, T. 1969. Sapid components in carrot. J. Food Sci. 39:392. Phan, C.T., Hsu, B., and Sarkar, S.K. 1973. Physical and c chemical changes occuring in the carrot root during storage. Can. J. Plant Sci. 53: 635. Powel, J. 1981. Private Communication. Economic and Statistical Analysis Div., USDA. Riddle, P.J. and MacGillivary, J.H. 1966. Relation of dry matter to soluable Solids in carrots and peppers. Proc. Amer. Soc. Hort. Sci. 89:381. Ross, I.J. 1975. Humidity. in "Instrumentation and Measurement for zenvironmental Sciences" 8-01, Amer. Soc. Agr. Eng., St. Joseph,MI. 161 Rummel, R. J. 196?. Understanding Factor Analysis. Conflict Resolution. 11:hhfl. Rygg, G.L. 19N5. Sugars in the root of carrot. Plant Physiol. 20:”?. Ryhage,R. and von Sydow,E. 1963. Mass spectrometery of terpenes. I. Monoterpene hydrocarbons. Acta Chem. Scand. 17: 2025. Sarkar, S.K. and Phan, O.T. 1979. Naturally-Occuring and Ethyleen-Induced Phenolic Compounds in the Carrot Root. J. of Food Prot. h2(6):526. Scheerens, J.C., and Hosfield, G.L. 1976. The feasibility of improving eating quality of table carrots by selecting for total soluble solids. J.Amer. Soc. Hort. Sci. 101(6):?05. Seifert, R.M., Buttery,R.G., and Ling, L.C. 1968. The composition of carrot seed oil. J. Sci. Food Agr. 19:383 Seifert, R.M., and Buttery, R.C. 1978. Characterization of Some Previously Unidentified Sesquiterpenes in Carrot Roots. J. Agric. Food Chem. 26(1):181. Sell, H.M. 1956. Occurrence of a bitter principle in carrots. Science 128:98M. Simon, P.W. 1981. Private Communication. Univ. of Wisconsin. Simon, P.W., Lindsay, R.C., and Peterson, C.E. 1980a. Analysis of Carrot Volatiles Collected on Porous Polymer Traps. J. Agr. Food Chem. 28(3):5h9. Simon, P.W., Peterson, C.E., and Lindsay, R.C. 1980b. Correlations between Sensory and Objective Parameters of Carrot Flavor. J. Agr. Food Chem. 28(3):559. Simon, P.W., Peterson, C.E., and Lindsay, R.C. 1980c. Genetic and Environmental Influences on Carrot Flavor. J. Amer. Soc. Hort. Sci. 105(3):h16. Sondheimer, E. 1956. Bitter Flavor in Carrots. III. The isolation and identification of a compound with spectral characteristics similar to hydrocarbon extracts of bitter carrots. Food Res. 30:296. Sondheimer, E. 1957. The isolation and identification of 3-Methyl-6-methoxy-8-hydroxy-3,h-dihydroisocoumarin from carrots. J. Amr. Chem. Soc. 79:5036. 162 Sondheimer, E., Phillips, W.F., and Atkin, J.D. 1955. Bitter Flavor in Carrots. I. A Tentative Spectrophotometric Method for the Estimation of Bitterness. Food Res. 30:659. Stuiver M.,. Doctoral Thesis, Rijks Univ., Groningen, Holland, 1958. Quoted by Devries H. and Stuiver, M. 1961. In "Sensory Communication" (Rosenblith W.A., ed.) Wiley, New York,N.Y. p159. Teranishi, R., Issenberg, P., Hornstein, I., and Wick, E.L. 1971. Flavor Research Principles and Techniques. Marcel Dekker, Inc., New York. 1971. Tukey, J.W. 1953. "The problem of multiple comparisons." Mimeographed. Prinnceton Univ., Math. Dept., -Princeton, N.J. . Van den Berg, L., and Lentz, C.P. 1966. Effect of temperature relative humidity, and Atmospheric composition on changes in quality of carrots during storage. Food Tech. July. Van den Berg, L., and Lentz, C.P. 1973. High humidity Storage of Carrots, Parsnips, Rutabagas, and Cabbage. J. Amer. Soc. Hort. Sci. 98(2):129. Warner, R., Ernst, J.O., Boundy, B.K., and Evans, C.D. 197H. Computer Handling of Taste Panel Data. Food Tech. Novzh2. Weichman, J. 1977. Physiological Responce of Root Crops to Controlled Atmospheres. Acta Hort. 62:122 Withycombe, D.E., Mookherjee, B.D., and Hruza, A. 1978. Isolation of trace volatile constituents of hydrolized vegetable protein via porous polymer headspace entrainment. In "Analysis of Foods and Beverages Headspace Techniques" p.81. Academic Press. New York, NY.