Discussion
 
Although other studies with microscale fermentation have shown results using much smaller volumes (1L), our contribution focuses in the relevance of volume fermentation size even when bigger volumes are considered. With regards to repeatability, all tank sizes proved to be adequate, since CV values were in general low. Certain trend was observed to have increased variability in 25 and 50 L tanks, but differences were small and the CVs obtained were very satisfactory (usually below 15% when determining phenolics and 5% on grape and wine composition). This is a very important result, since one of the main concerns of researchers in viticulture and oenology is that reducing tank size in their experiments can increase variability in the fermentation stage, and therefore obtaining less reliable results. According to our data, decreasing tank size from 100 to 10 L does not cause an increase in variability and, therefore, the reliability of the results is very good.
 
However, having similar reliability in terms of variability does not mean that tank size does not affect wine representativeness. For both varieties, we observed that the biggest volume was more representative of commercial scale fermentation, particularly for anthocyanins (first component in PCA). Thus, 10 L tanks achieved the lowest concentration of anthocyanin and phenol extraction into the wine, with the extraction of non-acylated anthocyanins being benefited. De Villiers et al. (2004) found that non-acylated glycosides are more easily extracted, followed by acetyl glycosides and p-coumaroylated, being the latter the more difficult to extract from grape to wine. On the other side, procyanidins, included majorly in the second component of PCA, were extracted in larger quantities in the commercial size-tank, although 10 L, 25 L and 100 L showed similar scores for this component compared to the commercial scale wine. The pump-overs and the extended maceration made in the commercial wine may affect differently than the gently hand punched down made in the small-scale, due to an additional mechanical action of the pump which is not applicable to small volumes, and leads to a much greater concentration of monomers into the wine. However, despite different extraction of monomers, dimers and trimers, the total procyanidin content is much similar between tanks than that observed in the extraction of anthocyanins.
 
Tank size affected fermentation dynamics in both varieties, the effects being clearer in CS tanks, where fermentation took place more slowly due to the smaller berry size type variety. In both varieties, the smallest tank (10 L) fermented the fastest, no differences being found between the remaining 3 sizes in TE, and being gradually slower as tank size increased in CS. However, tank size did not affect the total time required to complete fermentation (Figure 1).
 
 
All in all, according to our results, the smallest tank size used in this study could be representative enough when the goal of winemaking is to compare different field or winemaking strategies (i.e., viticulture practices or yeast trials), as variability was not affected by tank size. Nevertheless, when the objective of small scale winemaking is to perform the wine extraction and phenolic composition, mainly for red phenolic varieties, increasing tank volume (up to 100 L) would be necessary to obtain comparable results to commercial scale wines.
 
 
 
Conclusions
 
Small-scale winemaking represents a necessary and valuable tool for viticulture and oenological research, although small size tanks should be only used when the objective of research is to compare different field or winery treatments in relative terms. Instead, to define the phenolic composition of a commercial wine, bigger volumes are needed to produce wines with similar properties and phenolic content.