Thesis (MSc Food Sc )--Stellenbosch University, 2000. / ENGLISH ABSTRACT: This study shows Fourier transform near infrared (FT-NIR) spectroscopy's application on wine, fruit and dried fruit for quantitative determinations or as a discriminative method for classification purposes. During wine production optimum yeast growth, resulting in healthy alcohol fermentation rates, is monitored by the amount of free amino nitrogen (FAN) present in the must. The status of malolactic fermentation (MLF) in Chardonnay wines is monitored by determining the degree of conversion of malic to lactic acid. Ethyl carbamate (EC), a suspected carcinogen, is mainly formed during ageing of wine and is restricted by legislation in some countries. It is therefore necessary to determine the EC content in wine. Fourier transform near infrared spectroscopy can be used on wine as a rapid method to measure the °Brix (residual sugars) content of must and to discriminate between different must samples in terms of their FAN values. It can also be used as a rapid method to discriminate between Chardonnay wine samples in terms of the MLF status and between table wine samples in terms of the EC content. Calibrations were derived and it was found that a very strong correlation existed in the sample set for the FT-NIR spectroscopic predictions for °Brix (r = 0.99, SEP = 0.31%), but poorer correlations for the FAN (r = 0.405, SEP = 275%), malic acid (r = 0.64, SEP = 1.02%), lactic acid (r = 0.61, SEP = 1.35%) and EC predictions (r = 0.47, SEP = 3.6%). When soft independent modelling by class analogy (SIMCA) was applied as a discriminative method, the must and wine samples were classified in terms of their FAN and EC values and MLF status, respectively, obtaining results with recognition rates exceeding 80%. The canning of peaches has become a significant industry in South Africa, producing approximately 4.1 million cartons per year. Fourier transform near infrared spectroscopy was applied as an alternative non-destructive method for the quantitative determination of the total soluble solid (TSS) content of whole fresh peaches. The TSS content of fresh clingstone peaches is an indication of the internal quality, maturity and perceived sweetness of the fruit for the peach canning industry. By determining the TSS, fresh peaches can be graded and the farmers compensated accordingly. Results obtained by building QUANT+™calibrations for the TSS content (r = 0.96, SEP = 0.55%) showed acceptable accuracy and can replace the present destructive methods. Peak periods during the harvesting season necessitate storage of peaches for up to three weeks before canning. Approximately 5 - 10% of the peaches stored, disintegrate during canning due to loose skin, large stone cavities, soft flesh and rot. The storage potential of fresh clingstone peaches can be successfully predicted with FT-NIR and SIMCA models, using subjective internal quality evaluations. Results with recognition rates exceeding 80% were obtained in most cases and this method proved useful as a non-destructive method of quality assessment. By applying this method, losses caused when storing peaches with poor storage quality will be reduced. The golden sultana industry plays an important role in the dried fruit exporting market of South Africa. Due to the large numbers of consignments that must be checked upon arrival, and the need for rapid decision making during processing, it is essential to replace the present time-consuming analytical methods. Fourier transform near infrared spectroscopy was used as a rapid, analytical technique to determine whether the S0₂ and moisture contents of sultanas are within specifications upon arrival at the factory and during processing. High positive correlation was found between the measured values and those predicted by FT-NIR spectroscopy for S0₂ (r = 0.99, SEP = 24.09%) and moisture (r = 0.99, SEP = 0.051 %) contents. / AFRIKAANSE OPSOMMING: Hierdie studie dui op Fourier transformasie naby-infrarooi (FT-NIR) spektroskopie se toepassing op wyn, vrugte en droëvrugte vir die uitvoer van kwantitatiewe bepalings of vir klassifikasie doeleindes om as 'n diskriminasie metode te dien. Gedurende wynproduksie word die optimum groei van giste wat lei tot 'n gesonde alkohol fermentasie gemonitor deur die hoeveelheid vry-aminostikstof (VAS) wat in die mos teenwoordig is te bepaal. Die status van appelmelksuurgisting (AMG) in Chardonnay wyne word gemonitor deur die mate van omskakeling van appelsuur na melksuur te bepaal. Etielkarbamaat (EK), 'n vermoede karsinogeen wat hoofsaaklik in verouderde wyne voorkom, word in sekere lande deur wetgewing beperk en dus die bepaling van die EK inhoud van wyne noodsaak. Fourier transformasie naby-infrarooi spektroskopie kan op mos toegepas word as 'n vinnige metode vir die bepaling van °Brix (residuele suiker) en om tussen die monsters te onderskei in terme van hulle VAS inhoud. FT-NIR kan ook gebuik word as 'n vinnige metode om tussen Chardonnay monsters te onderskei op grond van die status van AMG en tussen tafelwyn monsters op grond van die EK inhoud. Kalibrasies is ontwikkel en daar is gevind dat baie sterk korrelasies bestaan in die monsterstel vir die FT-NIR spektroskopiese voorspelling van °Brix (r = 0.99, SEP = 0.31%), maar swakker korrelasies vir die VAS (r = 0.405, SEP = 275%), appelsuur (r = 0.64, SEP = 1.02%), melksuur (r = 0.61, SEP = 1.35%) en EK voorspellings (r = 0.47, SEP = 3.6%). Met die toepassing van soft independent modelling by class analogy (SIMCA) as diskriminasie metode, is die mos- en wynmonsters geklassifiseer op grond van hul VAS en EK waardes en die status van AMG, en herkenningswaardes van bo 80% is onderskeidelik behaal. Die inmaak van perskes het 'n beduidende industrie in Suid-Afrika geword en produseer jaarliks ongeveer 4.1 miljoen kartonne. Fourier transformasie nabyinfrarooi spektroskopie is toegepas as alternatiewe, nie-beskadigende metode om kwantitatiewe bepalings van die totale oplosbare vastestowwe (TOV) inhoud van heel vars perskes, te doen. Vir die perske inmaak-industrie is die TOV inhoud van vars taaipitperskes 'n aanduiding van interne kwaliteit, rypheid en die soetheid van die vrugte. Vars perskes kan gradeer word deur die TOS te bepaal en sodoende kan boere oreenkomstige vergoeding ontvang. Resultate' wat verkry is deur QUANT+™ kalibrasies vir TOS inhoud te ontwikkel (r = 0.96, SEP = 0.55%), dui op aanvaarbare akkuraatheid en kan die huidige metodes vervang. Tydens oestyd kom piektye voor wanneer dit soms nodig is om perskes vir tot drie weke op te berg voordat dit ingemaak kan word. Ongeveer 5 tot 15% van hierdie opgebergte perskes disintegreer tydens inmaak omdat opberging lei tot defekte in die perskes soos skille wat loskom, groot pitholtes, sagte vleis en vrot. Die opbergingspotentiaal van vars taaipitperskes kan suksesvol voorspel word deur FTNIR en SIMCA modelle te bou en subjektiewe interne kwaliteitsevaluerings daarop toe te pas. Herkenningsresultate wat 80% in die meeste gevalle oorskry, is behaal wat hierdie metode as 'n suksesvolle nie-beskadigende kwaliteitbepalingsmetode bewys. Hierdie metode sal verliese kan beperk wat voorkom as gevolg van opberging van perskes met swak opbergingskwaliteit. Die goue sultana industrie speel 'n belangrike rol in die droë vrugte uitvoermark van Suid-Afrika. As gevolg van die hoeveelheid monsters wat gelyktydig getoets moet word en besluite wat vinnig geneem moet word tydens prosessering, is dit belangrik om die huidige tydrowende analitiese metodes te vervang. Fourier transformasie naby-infrarooi spektroskopie is gebruik as 'n vinnige, analitiese tegniek om tydens ontvangs by die fabriek en gedurende prosessering te bepaal of die S0₂- en voginhoud van goue sultanas binne die spesifikasies val. Goeie positiewe korrelasie is gevind tussen die bepaalde en voorspelde FT-NIR spektroskopiese waardes vir SO₂- (r = 0.99, SEP = 24.09%) en voginhoud (r = 0.99, SEP = 0.051 %).