The aromatic profiles of four strains of Brevibacterium linens, one strain of Brevibacterium sp. and one strain of Microbacterium sp. were determined with some pure cultures of these microorganisms in standard trypcase soy liquid medium, which enabled four of these six strains to produce flavour compounds of ripened cheese. Thirty-two flavour compounds were identified by gas chromatography and mass spectrometry. The identified flavour compounds included the following: fatty acids, alcohols, methylketones, sulphur compounds, aromatic compounds and a pyrazine. Some important differences were found among the six strains studied. The four strains of B. linens had similar flavour profiles. Their typical flavour was probably due to dimethyltrisulphide. The two other strains did not appear to produce this compound. Three strains produced significant amounts of the floral aromas phenylethanol and phenylpropanone.

In the study of the nitrogen influence on alcoholic fermentation with simulated oenological conditions, it is necessary to have objective and quantifiable criteria to compare different fermentations and to estimate the nitrogen effect on the course of the reaction. Much work has already been performed on fermentation models, and many relations have been proposed; unfortunately, they are usually only applicable with the conditions used by their authors. Among these models, we especially can mention one proposed by Ludeking and Piret' which links the production to the growth. However, according to Aiyar and Ludeking,2 it is only utilizable for initial sugar concentrations below 20 g/L in the case of alcoholic fermentation.Other authors propose relations between the specific rate of production and the ethanol c~ncentration,~'~ but they use the concentration of added ethanol and not the ethanol produced, whereas Novak et al.5 have shown that these two types of ethanol have very different effects on the biomass. Some propose relations where the substrate consumption is taken into account.677 Recently, Luong' proposed a model applicable to a great number of fermentations found in the literature, but not corresponding with the experimental curves we obtained.Concerning models peculiar to oenology, it is important to mention Holzberg and co-workers,' who propose two relations based on the exponential or the stationary phases of growth, respectively. Boulton" gives a predictive model, including substrate concentration, time, and required temperature, but omitting important factors as pH, the level of bisulfite ions, and especially the nitrogen concentration, which particularly preoccupies the current authors.None of these models was satisfactory with regard to the fermentation curves described in the present communication. So a new model was developed, giving a good description of experimental results-a model in which nitrogen was taken into account. This communication describes such a model, its validity, and especially its interest in a study on nitrogen influence.