To be feasible the use of DRIS to assess the nutritional status of plants, the first step is establish the DRIS norms or standard. The DRIS norms consist on average and standard deviation of dual ratio between nutrients (N/P, P/N, N/K, K/N, etc.) obtained from a crop reference population (Table 1), but, it is necessary that the crop reference shows high yield (Beaufils, 1973). This method has been followed along the years (Jones, 1981; Alvarez V. & Leite, 1999; Silva et al., 2009; Maccray et al., 2010; Serra et al., 2010a, b; Serra et al., 2012).

The data bank to compose the DRIS norms is formed by the crop yield and chemical analysis of leaf tissue, and this information can be obtained from commercial crop or experimental units. The size of the data bank is not a factor that is directly related to the quality of the DRIS norms (Walworth et al., 1988; Sumner, 1977).

Walworth et al. (1988) observed that, when they used 10 data to establish the DRIS norms, the results obtained were more accurate then the use of a large number of data. What is more important to improve efficiency on DRIS norms is the quality of the data, because it is not accepting the use of sick plants to compose the data bank to establish the DRIS norms.

To make part of the DRIS norms, the rations between nutrients can be selected by the direct form (N/P) or reverse (P/N), but, there is more than one way to change the ratio that is going to compose the DRIS norms. Bataglia et al. (1990) used the entire dual ratio without selecting the direct or reverse form, and other researchers used the transformation by natural log (Beverly, 1987; Urano et al., 2006, 2007; Serra et al., 2010a, b; Serra et al., 2012).

With many ways to select the ratio to compose the DRIS norms there is a necessity to establish the most efficiency way for each crop that results in a better efficiency of the system. Silva et al. (2009) tested the dual ratio selection using the "F" value (Jones, 1981; Letzsch, 1985; Walworth & Sumner, 1987) and "r" value (Nick, 1998) in cotton crop, on his turn, Silva et al. (2009) did not test the criterion of choice the ratio by log transformation or the use of all nutrient ratio as it were made by Alvarez V. & Leite (1999) and Serra et al. (2010a, b).

Results obtained by Serra et al. (2012) showed that the use of "F" value or log transformation in nutrient ratio to define the norms produced different DRIS index, furthermore, when the DRIS index is interpret by Beaufils ranges the difference observed among index was reduced, showed less difference between the two groups of norms.

Following the premises of DRIS proposed by Beaufils (1973), it is feasible to change the dual ratio (A/B or B/A) that is more important to compose the DRIS norms. This way it is expected that the dual ratio from crop with high-yielding (reference population), composed with healthy plants, shows less variation than the population of plants with low-yielding (non-reference population), thus, the relation between variance ratio method, the F value, was defined as the variance ratio of low-yielding (non-reference) and high-yielding population (reference), and the order of the ratio with the highest value was chosen among the variance ratios (Jones, 1981; Letzsch, 1985; Walworth & Sumner, 1987).

The utilization of the relationship between variance ratio method ("F" value) from low- yielding and high-yielding is the most used method to define the DRIS norms. The method "F" value is defined on the data bank divided into two groups (non-reference and reference), and the choice of ratio directly (A/B) or inverse (B/A) defined by relationship between variances from the two populations, in which the ratio chosen will result arises from the following analysis (Jones, 1981; Letzsch, 1985; Walworth & Sumner, 1987):

Then: the dual ratio that will make part of the DRIS norms will be A/B, on the another it will be B/A. S2 is the variance of the dual ratio of the reference population and non-reference.

Besides the selection of forward or reverse ratio to compose the DRIS norms, the same principle can be selected with regard to the significance of F value, which can be 1%, 5% or 10% (Wadt, 1999), and feasible to use all dual ratio, which was selected by the largest ratio of variances, without the rigour of significance (Beaufils, 1973; Jones, 1981; Walworth & Sumner, 1987; Serra, 2011).

One can observe on literature that there are not any consensus about which methodology is more efficient to use. Jones (1981) did not select for significance, but he selected by the biggest reason of variances, as well as Raghupathi et al. (2005); Guindani et al. (2009); Sema et al. (2010); Serra et al. (2012). However, Wadt (2005) used the "F" value for the selection of dual ratio with a significance of 10%, excluding from the norms the dual ratio that was with significance above this value.

When selecting the dual ratio by significance of the "F" value, the sum of DRIS indexes does not give a zero value, in this case some nutrients can remain with a larger number of dual ratio than those with fewer ratios. However, Wadt et al. (1999) concludes that the rigour of the selection by the significance of "F" value generates greater efficiency for the diagnosis, in studies made with coffee crop (Coffea canephora Pierre).